Edited by Katsuyoshi Kondoh

Powder Metallurgy
Edited by Katsuyoshi Kondoh

Published by InTech
Janeza Trdine 9, 51000 Rijeka, Croatia

Copyright © 2012 InTech
All chapters are Open Access distributed under the Creative Commons Attribution 3.0
license, which allows users to download, copy and build upon published articles even for
commercial purposes, as long as the author and publisher are properly credited, which
ensures maximum dissemination and a wider impact of our publications. After this work
has been published by InTech, authors have the right to republish it, in whole or part, in
any publication of which they are the author, and to make other personal use of the
work. Any republication, referencing or personal use of the work must explicitly identify
the original source.

As for readers, this license allows users to download, copy and build upon published
chapters even for commercial purposes, as long as the author and publisher are properly
credited, which ensures maximum dissemination and a wider impact of our publications.

Statements and opinions expressed in the chapters are these of the individual contributors
and not necessarily those of the editors or publisher. No responsibility is accepted for the
accuracy of information contained in the published chapters. The publisher assumes no
responsibility for any damage or injury to persons or property arising out of the use of any
materials, instructions, methods or ideas contained in the book.

Publishing Process Manager Daria Nahtigal
Technical Editor Teodora Smiljanic
Cover Designer InTech Design Team

First published February, 2012
Printed in Croatia

A free online edition of this book is available at www.intechopen.com
Additional hard copies can be obtained from orders@intechweb.org

Powder Metallurgy, Edited by Katsuyoshi Kondoh
p. cm.
ISBN 978-953-51-0071-3


Preface VII
Chapter 1 Selection of Best Formulation for Semi-Metallic
Brake Friction Materials Development 1
Talib Ria Jaafar, Mohmad Soib Selamat and Ramlan Kasiran
Chapter 2 Porous Metals and Metal Foams Made from Powders 31
Andrew Kennedy
Chapter 3 Aluminium Alloy Foams: Production and Properties 47
Isabel Duarte and Mónica Oliveira
Chapter 4 The Fabrication of Porous Barium Titanate
Ceramics via Pore-Forming Agents (PFAs)
for Thermistor and Sensor Applications 73
Burcu Ertuğ
Chapter 5 Hybrid Gas Atomization for Powder Production 99
Udo Fritsching and Volker Uhlenwinkel


From high-performance, economical and environmental points of view, Powder
metallurgy process shows remarkable advantages in production of parts and
components due to their special compositions by elemental mixing and 3-dimensional
near net shape forming methods. Powder metallurgy process can be applied to not
only metal materials but also ceramics and organic materials, which both are
employed as structural and electrical products. Author contributions to Powder
metallurgy present excellent and significantly important research topics to evaluate
various properties and performance of P/M materials for applying these materials as
actual components. In particular, the life estimation of P/M ferrous materials by
sliding contact fatigue test and tribological performance evaluation of P/M semi-
metallic materials are focused and introduced in this book.

Katsuyoshi Kondoh
Joining and Welding Research Institute (JWRI)
Osaka University

Selection of Best Formulation for Semi-Metallic
Brake Friction Materials Development
Talib Ria Jaafar
, Mohmad Soib Selamat
and Ramlan Kasiran

Advanced Materials Centre, SIRIM Berhad, 34, Jalan Hi-Tech 2/3,
Kulim Hi-Tech Park, Kulim,
Faculty of Mechanical Engineering, University Technology MARA, ShahAlam
1. Introduction
Brake friction materials play an important role in braking system. They convert the kinetic
energy of a moving car to thermal energy by friction during braking process. The ideal
brake friction material should have constant coefficient of friction under various operating
conditions such as applied loads, temperature, speeds, mode of braking and in dry or wet
conditions so as to maintain the braking characteristics of a vehicle. Besides, it should also
posses various desirable properties such as resistance to heat, water and oil, has low wear
rate and high thermal stability, exhibits low noise, and does not damage the brake disc.
However, it is practically impossible to have all these desired properties. Therefore, some
requirements have to be compromised in order to achieve some other requirements. In
general, each formulation of friction material has its own unique frictional behaviours and
wear-resistance characteristics.
Friction material is a heterogeneous material and is composed of a few elements and each
element has its own function such as to improve friction property at low and high
temperature, increase strength and rigidity, prolong life, reduce porosity, and reduce noise.
Changes in element types or weight percentage of the elements in the formulation may
change the physical, mechanical and chemical properties of the brake friction materials to be
developed (Lu, 2006; Cho et Al., 2005; Mutlu et al., 2006 & Jang et al., 2004). Earlier
researchers have concluded that there is no simple correlation between friction and wear
properties of a friction material with the physical and mechanical properties (Tanaka et al.,
1973; Todorovic, 1987; Hsu et al. 1997 & Talib et. al, 2006). Therefore, each new formulation
developed needs to be subjected to a series of tests to evaluate its friction and wear
properties using brake dynamometer as well as on-road braking performance test to ensure
that the brake friction material developed will comply with the minimum requirements of
its intended application.
Two major types of brake dynamometers are commonly used to evaluate the friction and
wear characteristics of the friction materials are the inertia dynamometer and CHASE
dynamometer. Inertia dynamometer is used to evaluate a full size brake lining material or
brake system by simulating vehicles braking process but it is time consuming and more

Powder Metallurgy 2
expensive. On a smaller scale, CHASE dynamometer features low capital expenditure and
shorter test time (Tsang, 1985). Chase machine uses a small sample of friction material with
a size of 1 inch x 1 inch x 0.25 inch. These brake dynamometers has been used to tests
friction materials for quality control, lining development and friction materials property
assessments in a lab scale rather then having a series of vehicle tests on a test track or road
(Sander, 2001).
The two main types of tests used to evaluate the performance under different loading,
speed, temperature and pedal force are, namely, inertia-dynamometer and vehicle-level
testing. Inertia-dynamometer test procedures or vehicle testing simulation is used as a cost-
effective method to evaluate brake performance in a laboratory-controlled environment. The
automotive industry uses inertia-dynamometer testing for screening, development and
regular audit testing. Blau postulated that there is no laboratory wear test of vehicle brake
materials can simulate all aspects of a brake’s operating environment (Blau, 2001). Vehicle
testing on the test track is the ultimate judge for overall brake performance testing and
Generally, in normal life we cannot avoid friction phenomenon. It still happens as long as
there is a relative motion between two components. Even though friction can cause wear of
materials, sometimes the process of friction is required such as in the brake system, clutch,
and grinding. During a braking process, brake pads or brake shoes are pressed against the
rotating brake disc or drum. During this process the friction materials and the brake disc are
subjected to wear.
Friction is a continuous process but wear is a more complicated process than friction
because it involves plastic deformation plus localised fracture event (Rigney, 1997),
microstructural changes (Talib et al., 2003), and chemical changes (Jacko, 1977). Wear
process in dry sliding contacts begins with particle detachment from the contact material
surface due to formation of plastic deformation, material transfers to the opposite mating
surface and formation of mechanical alloyed layers (Chen & Rigney 1985), finally
elimination of wear fragments from the tribosystem as the wear debris. Wear mechanism in
the operation during braking is a complex mechanism and no single mechanism was found to
be fully operating (Rhee, 1973 & 1976; Bros & Sciesczka, 1977; Jacko et al., 1984; Talib et. al.
2007) and the major wear phenomena observed during braking processes were; (i) abrasive (ii)
adhesive (iii) fatigue (iv) delamination and (v) thermal wear.
Friction and wear characteristics of friction material play an important role in deciding
which new formulations developed are suitable for the brake system. The friction and wear
behaviours of automotive brake pads are very complex to predict which depend on the
various parameters such as microchemical structure of the pad and the metallic counter-
face, rotating speed, pressure and contact surface temperature (Ingo et al., 2004).
Composition and formulation of brake pads also play a big role on the friction behaviour,
and since composition-property relationship are not known well enough, the formulation
task is based on trial and error and thus is expensive and time consuming (Österl & Urban,
2004). Generally brake pads have a friction coefficient, µ between 0.3 and 0.6 (Blau, 2001).
In this work, ten (10) new friction material formulations which are composed of between
eight (8) to foureen (14) elements have been developed using power metallurgy technique.
In addition, a commercially-available brake pad, labelled as COM, was chosen for

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 3
comparison purposed. Each sample was subjected to density, hardness, porosity, friction
and wear, brake effectiveness and on-road braking performance tests in accordance with
various relevant international standards. The best formulation was selected based on the
following methodology;
i. First screenings - screening of the developed formulations based on the results of the
physical and mechanical tests.
ii. Second screening – screening of the developed formulations based on the results of
friction and wear tests performed on CHASE brake lining friction machine.
iii. Third screening - screening of the developed formulations based on the results of brake
dynamometer tests.
iv. Final selection - selection of the best developed formulations is based on the compliance
with the on-road braking performance requirements.
Correlation among the mechanical, tribological and performance will also be discussed in
this work. Wear phenomena on the worn surface after on-the road performance test will be
examined and postulated.
2. Materials and method
2.1 Semi-metallic friction materials
Ten semi-metallic brake pad formulations which composed of between eight (8) to fourteen
(14) ingredients were produced in this study using powder metallurgy route (Table 1). The
powder metallurgy route consists of the following processes, namely, (ii) dry mixing, (ii)
preparation of backing plate, (iii) pre-form compaction (iv) hot compaction, (v) post-baking,
and (vi) finishing. The prototype samples were marked as SM1, SM2, SM3, SM4, SM5, SM6,
SM7, SM8, SM9 and SM10. Figure 1 shows two (2) example microstructure of the newly
developed semi-metallic brake pad. It can be seen that the brake pads developed are not a
homogenous material. The particle size of each element is not unifrom in size and the
distribution of the element is not well dispersed in the matrix.

(a) (b)
Fig. 1. Surface morphology of semi-metallic brake pad; (a) sample SM1, (b) sample SM4

Powder Metallurgy 4
Formulation (Weight %)
SM1 SM2 SM3 SM4 SM5 SM6 SM7 SM8 SM9 SM10
10.0 10.0 9.0 9.0 9.0 9.0 9.0 8.0 12.0 9.0
- - 2.0 - 2.0 2.0 - - - 3.0
Steel fiber
20.0 23.0 31.0 20.0 30.0 31.0 22.0 24.0 22.0 25.0
Organic fiber
5.0 - - 10.0 2.0 2.0 10.0 8.0 7.0 5.0
Copper fiber
- - 2.0 6.0 2.0 3.0 6.0 - 8.0 3.0
16.0 19.0 7.0 13.0 15.0 7.0 13.0 11.0 16.0 6.0
- - 3.0 - 3.0 3.0 - - 5.0
Iron oxide
34.0 24.0 18.0 9.0 16.0 18.0 9.0 15.0 3.0 21.0
Novacite silica
- 3.0 - 3.0 - - 3.0 2.0 6.0 3.0
Alumina oxide
- 2.0 1.0 5.0 2.0 1.0 2.0 2.0 2.0
Zinc oxide
1.0 - 1.0 - 2.0 - - 3.0 - 2.0
- 3.0 4.0 3.0 3.0 3.0 3.0 3.0 3.0 5.0
White rock
- - 2.0 3.0 2.0 3.0 3.0 6.0 3.0 3.0
8.0 10.0 20.0 19.0 8.0 18.0 20.0 14.0 - 7.0
Friction dust
6.0 6.0 - - 4.0 - - 4.0 15.0 6.0
100 100 100 100 100 100 100 100 100 100
Table 1. Ingredients of semi-metallic brake pad
2.2 Physical and mechanical tests
Each sample produced is subjected to density, porosity and hardness tests. Density of semi-
metallic brake pads was obtained using Archimedes’ principle in accordance with
Malaysian Standard MS 474: Part 1: 2003 test procedures. Hardness was measured using a
Rockwell hardness tester model Mitutoyo Ark 600 in S scale in accordance with Japanese
Industrial Standard JIS 4421: 1996 test procedures. The hardness of the samples is the
arithmetic mean of ten measurements. Porosity was obtained in accordance with JIS 4418:
1996 test procedures using a hot bath model Tech-Lab Digital Heating.
2.3 Friction and wear tests
Friction coefficient and wear were results were obtained which is in compliance with Society
of Automotive Engineer SAE J661 test procedures. In this test, the sample was pressed
against a rotating brake drum with a constant rotating speed of 417 rpm under the load of
647 N and subjected to test program as shows in Table 2. Briefly, each sample was subjected
to seven test runs with the following sequences; (i) baseline, (ii) first fade, (iii) first recovery,
(iv) wear, (v) second fade, (vi) second recovery, and (viii) baseline rerun. The samples
thickness were measured and weighed before and after testing. Friction coefficient and wear
tests were conducted by Greening Testing Laboratories Inc., USA using CHASE machine.

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 5
Test Sequence Load
Braking mode
Conditioning 440 312 Continuous braking for 20 mins
Initial thickness & mass
222 208
Continuous braking for 5 mins
Baseline run
667 417
Intermittent braking
10 s ON, 20 s OFF for 20 applications
First fade run 667

10 minutes or until 288
C is attained
which ever come first
First recovery run
667 417
10 seconds application at 260, 204 , 149
and 93
Wear run
667 417
20 s ON, 10 s OFF for 100 applications.
Second fade run
667 417
10 mins or until 343
C is attained
which ever come first
Second recovery run
667 417
10 seconds application at 316, 260, 204 ,
149, 93
Baseline rerun
667 417
10 s ON, 20 s OFF for 20 applications
Final thickness and
mass measurement

Repeat initial thickness and mass
Table 2. Friction and Wear Assessment Test Program
2.4 Brake effective tests
The braking performance of the developed semi-material brake pads were determined using
brake dynamometer test in accordance with Society of Automotive Engineers standard SAE
J2552 issued in August 1999 (available from SAE, 400 Commonwealth Dr, Warrendale, PA
15096, USA). This standard assesses the effectiveness behavior of a friction material with
regard to pressure, temperature and speed. Vehicle brake simulations are conducted on an
inertia dynamometer, which simulate kinetic energy of the vehicle mass moving at speed.
Before the beginning of performance measurement, a burnishing period for conditioning the
lining/counterface pairs require more than 200 conditioning stops. After conditioning,
dynamometer-based lining tests were subjected to pressure-sensitive stops, speed-sensitive
drags, fade and recovery tests. Table 3, briefly shows the test sequences. Dynamometer
global brake effectiveness tests were conducted by Greening Testing Laboratories Inc., USA
using single end brake brake dynamometer. Each sample was conducted on a new brake
rotor. In this study, the focus is only on the friction coefficient and wear characteristics of the
The purpose of this investigation was to evaluate the performance of the semi-metallic brake
pads for Proton WIRA using brake inertia dynamometer. Figure 2 shows prototype brake
pad and Proton Wira’s drive shaft assembly. The technical specification of the brake
effectiveness test is shown in Table 4.

Powder Metallurgy 6

Bil Snub Cycle Speed
temp (
1. Green μ 30 1 80 to 30 3000 < 100
2. Burnish 32 6 80 to 30 Varying
< 100
3. Characteristic 1 6 1 80 to 30 3000 < 100
4. Speed/press
40 to 5
80 to 40
120 to 80
160 to 130
200 to 170
1000 to 8000
< 100
5. Characteristic 2 6 1 80 to 30 3000 < 100
6. Cold 1 1 40 to 5 3000 < 40
7. Motorway
100 to 5
0.9 V
0.5 V

0.6 g < 50
8. Characteristic 3 18 1 80 to 30 3000 < 100
9. Fade 1 15 1 100 to 5 16000
0.4 g
< 100
< 550
10. Recovery 1 18 1 80 to 30 3000 < 100
11. Temp/press
Sensitivity 100/80

8 1 80 to 30 Increasing
1000 to 8000
< 100

Ssnsitivity 500/300

9 1 80 to 30 3000

< 100

Pressure line
8 1 80 to 30 Increasing
1000 to 8000
< 550
13. Recovery 2 18 1 80 to 30 3000 < 100
14. Fade 2 15 1 100 to 5 16000
0.4 g
< 100
< 550
15. Recovery 3 18 1 80 to 30 3000 < 100

Table 3. Brake dynamometer test sequence

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 7
Item Specification
Vehicle System Simulated 1996 Proton Wira 1.5 GL Front
Brake Configuration single piston, separate function disc brake
Piston Diameter 54 mm
Rotor Diameter x Thickness 236 x 18 mm
Rotor Mass (nominal) 3.7 kg
Rotor Effective Radius 95.88 mm
Axle Load 830 kg
Test Inertia 34.7 kg·m²
Static Loaded Radius / Rolling Radius 287.02 mm
Simulated Wheel Load 421 kg
Wheel Rotation : right hand
Table 4. Technical specifications of the brake effective test

Fig. 2. Semi-metallic brake pad and front axle brake system
2.5 On-road performance tests
In the road performance test, the brake pads were fitted to the brake system of PROTON
WIRA 1.5GL with the following test conditions: (i) unladen vehicle, (ii) disconnected engine,
(iii) tire inflated to the manufacturer’s specifications, (iv) the road was hard, level and dry,
(vi) the wind speed was below 5 m/s. The road performance test divided into three types,
namely; (i) cold effectiveness test, (ii) heat fade test, and (iii) recovery test.
The on-road braking performance tests on car were performed following closely procedure
described in the ECE R13, Annex 3. Modification on the procedure was necessary due to the
limitation of the test track conditions (Table 5). Real application testing of the friction
materials were carried out using Proton Wira 1.5 (Table 6). Figure 3 shows the test
equipment set-up.

Powder Metallurgy 8
Type-O: Cold brake
 Initial vehicle speed 120 km/h 100 km/h
 Brake pedal force 65 – 500 N 65 – 500 N
 Engine disconnected Yes Yes
 Average temperature 65 – 100
C 65 – 100
 Vehicle must be laden & unladen Yes 2 people

Type-1: Fade test
Heating procedure:
 Braking speed 120 - 60 km/h 100 - 50 km/h
 Brake pedal force Equivalent to 3 m/s

50% of Type-O
 No. of brake application 15 12
 Vehicle must be laden Yes 2 people
 Engine connected Yes Yes

Hot performance:
 Initial vehicle speed 120 km/h 100 km/h
 Brake pedal force Same force obtained
in Type-O test
Same force obtained
in Type-O test
 Engine disconnected Yes Yes

Recovery procedure:
 Braking speed 50 - 0 km/h 50 - 0 km/h
 Brake pedal force Equivalent to 3 m/s

50% of Type-O
 No. of stops with 1.5 km interval 4 4
 Vehicle must be laden Yes 2 people
 Engine connected Yes Yes

Recovery performance:
 Initial vehicle speed 120 km/h 100 km/h
 Brake pedal force Same force obtained
in Type-O test
Same force obtained
in Type-O test
 Engine disconnected Yes Yes

Table 5. Modified ECE R13 test procedure

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 9

Fig. 3. Test equipment set-up; (a) Dewetron DEWE-5000 system, (b) pressure sensor and
thermocouple, (c) GPS receiver installed on the roof (d) pedal force sensor.

Manufcaturer PROTON
Model Proton WIRA 1.5S
Engine capacity 1,468 cc
Max. Power 66 kW
Max Torque 126NM @ 3000 rpm
Gear system manual
Wheel size 175/70/R 13
Tire pressure 190 kPa
Table 6. Test car specifications
In the cold test, the test was conducted with the brake lining temperature below 100 ˚C prior
to each brake application and comprised of six brakings including familiarization. The brake
test was carried out at the initial vehicle speed of 100 km/hr. The test data such as vehicle
speed, lining temperature and braking distance were recorded using a brake measuring
system from Dewetron model DEWE5000. If the pedal force applied is more than 500 N for
wheel locking to occur, the brake pad is considered fail to comply with the requirements
and the next test (heat fade and recovery test) will not be conducted.
GPS reciever
Pedal force

Powder Metallurgy 10
Fade test is used to evaluate the brake performance under high brake material temperature.
Prior to this test, the service brake of the test car was heated by successively applying the
brake. The initial speed at beginning of this heating procedure was set at 100 km/hr and
speed at the end of braking was set at 50 km/hr with brake pedal force capable of
generating 50 % type-O deceleration. This process was repeated for 12 brake applications.
Upon completing this heating procedure, the test vehicle was accelerate to initial vehicle
speed of 100km and brake was applied using the same pedal force as in cold effective test of
that particular sample. Immediately, the recovery test was conducted with the following test
procedure; (a) make four stops from 50 km/hr with the same pedal force applied during
heating process of heat fade test. Immediately after each stop, accelerate the vehicle to 50
km/hr and make subsequent stop, (b) accelerate the test vehicle to a speed of 100km and
then brake pedal was applied with the same pedal force cold test.
2.6 Microstructural examination
The worn surface after on-road performance test were analyzed using scanning electron
microscope model Leo equipped with a Oxford energy-dispersive X-ray analyzer (EDX).The
samples for microstructural examination were cut from a real-size brake pad and coat with
platinum using sputter coater to ovoid charging effect during the analysis.
3. Results and discussion
3.1 Physical and mechanical properties
Specific gravity is the relative density of a substance compared to the density of pure water
and porosity is the percentage of pore volume with the bulk total volume. Hardness is a
measure of material resistance to plastic deformation. The specific gravity, porosity and
hardness properties depend on the ingredients and weight percentage used as well as the
manufacturing process parameters. Test results of physical and mechanical properties are
shown in Table 7 and Figure 4. Ideally, the highest specific gravity should give the lowest
porosity and highest hardness reading. But in friction materials this postulation does not
apply as shown in this investigation. For example, sample SM6 has the highest specific
gravity reading but does producing the highest hardness result. Highest hardness reading
was recorded by sample SM1, but this sample is not producing the highest density and the
lowest porosity reading (Table 7).
Figure 4 exhibits the correlation among the specific gravity, porosity and hardness. Figure
4a indicates that the specimens with high porosity tend to exhibit low specific gravity.
However, Figure 4b shows that there is no simple correlation between hardness with
porosity. Brake pad should have a certain amount of porosity to minimize the effect of water
and oil on the friction coefficient and to reduce the brake noise. Hardness should be
decreased as much as is feasible to increase performance stability and the steel fiber content
should be less than 7% in order to reduce rotor thickness variation (Sasaki, 1995). He also
found that increasing porosity by more than 10 % could reduce the brake noise. But if the
porosity too much, the hardness will be reduced resulting increase in wear rate of friction
material. Friction material is not a homogeneous material, when the indenter hits on the
metallic component the hardness will be higher, otherwise when it hits on polymeric
component the hardness will be lower (Talib et al., 2008). Thus, the hardness of the friction
material is not a representative of the bulk property.

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 11
Ideally the friction materials developed should have the best physical and mechanical
properties in order to get the best brake effective performance. In case of friction materials,
this phenomenon does not apply (Todorovic, 1987; Filip et al., 1995; Talib et al., 2006). The
physical and mechanical properties of friction material can not be predicted based on type
of ingredient used, particle size and shape, weight percentage of the ingredient. It also
depends on manufacturing process parameters such as powder mixing duration,
compaction pressure, compaction duration, degassing time, and post curing temperature
and time. Based on the above observations, the best formulation can not be selected using
physical and mechanical properties. The physical and mechanical properties could be used
to control the quality of the formulations that has been developed during the manufacturing
process. Consistent physical and mechanical properties of the same formulation reveal that
the friction material manufacturing process is in control.

Bil Sample Specific
1. COM 2.69 10.4 69.9
2. SM1 2.76 7.9 85.1
3. SM2 2.73 2.6 83.8
4. SM3 2.36 24.2 76.7
5. SM4 2.30 21.4 70.6
6. SM5 2.27 20.2 50.0
7. SM6 3.25 2.5 69.7
8. SM7 2.75 7.7 69.6
9. SM8 2.65 9.5 60.5
10. SM9 2.32 3.1 73.0
11. SM10 2.43 16.1 51.1
Table 7. Physical and mechanical test results
3.2 Friction and wear properties
Friction material is a heterogeneous material and composed of a few elements. Therefore,
the selection of material and weight percentage used in the friction formulation will
significantly affect the tribological behaviour of the brake pad [Hoyer et al.1999]. Society of
Automotive Engineer introducing two letter codes in classifying the friction material, where
first letter represents normal friction coefficient and the second letter represents hot friction
coefficient [SAE J886] as shown in Table 8. Normal friction coefficient is defined as average
of the four readings taken at 200, 250, 300 and 400ºF on the second fade curve. The hot
friction coefficient is defined as the average of the ten readings taken at 400 and 300ºF on the
first recovery; 450, 500, 550, 600 and 650ºF of the second fade; and 500, 400 and 300ºF of the
second recovery run. Figure 5 shows sample SM5 CHASE test results.

Powder Metallurgy 12

Fig. 4. The relationship between physical properties of friction material used in this study:
(a) specific gravity Vs. porosity, (b) hardness Vs. porosity.

Class code Coefficient of friction
C Below 0.15
D Over 0.15 – not over 0.25
E Over 0.25 – not over 0.35
F Over 0.35 – not over 0.45
G Over 0.45 – not over 0.55
H Over 0.55
Z unclassified
Table 8. SAE Recommended Practice J866 list for codes and associated friction coefficient
Friction and wear characteristics of friction material play an important role in deciding
which new formulations developed are suitable for the brake system designed for a
particular vehicle. CHASE test is used in a laboratory for screening of new material
formulations prior to inertia dynamometer tests based on friction and wear test results. In
deciding which samples are to be subjected to dynamometer tests, the following
requirements were set; (a) shall have normal friction coefficient of class E and above, or a hot
of class D and above, (b) shall have friction coefficient above 0.15 between 200 and 550

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 13
inclusive in second fade, or between 300 and 200
F during the secondary fade. These
requirements are in line the requirement set by Automotive Manufacturer Equipment
Companies Agency, USA.

Fig. 5. Friction coefficient characteristic of sample SM5
Test results of friction and wear assessment tests are shown in Table 9 and Figure 6.
Analysis of test results showed that all samples developed met with the requirements.
During braking, the accumulation of heat will cause high surface temperature on the brake
lining materials. The degradation of the polymer materials may cause brake fade in which the
friction is reduced as the temperature increased [Begelinger et al. 1973; Rhee 1971]. The ensuing
reduction of friction coefficient may also be explained by the shearing of the peak asperities
and formation of friction during braking process [Talib et al. 2007]. All hot friction
coefficients of the prototype samples reduced except sample SM8. This phenomenon may be
due to the formation of metallic layer [Talib, 2001; Scieszka, 1980; Lim et al., 1987] and

Powder Metallurgy 14
carbon layer [Begelinger et al., 1973; Zhigao & Xiaofie, 1991]. Sample SM1 and SM2 have
lower average thickness loss than the commercial sample. The other eight samples have
higher average thickness loss. Even though, higher wear loss resulted in shorter life, the
samples which higher average thickness loss will also be subjected to brake effective
dynamometer so that correlation thickness loss between these two tests could be made.

Normal Friction
Hot Friction Thickness
loss (mm)
 Code  Code
COM 0.422 F 0.385 F 2.79
SM1 0.385 F 0.316 E 0.51
SM2 0.450 F 0.383 F 0.76
SM3 0.459 G 0.352 F 1.52
SM4 0.471 G 0.373 F 2.79
SM5 0.457 G 0.360 F 1.02
SM6 0.417 F 0.345 E 2.03
SM7 0.438 F 0.430 F 3.31
SM8 0.532 G 0.544 G 6.09
SM9 0.374 F 0.322 E 1.27
SM10 0.554 H 0.458 G 5.84
Table 9. Friction and Wear Assessment Test Results
Figure 7 shows the relationship between the hardness with the friction coefficient and
average wear. Figure 7a indicates that the sample with high hardness tend to exhibit low
friction coefficient. More & Tagert (1952) and Mokhtar (1982) likewise concluded that the
coefficient of friction decreased with increase in hardness. Generally, the harder samples were
supposed to have a lower average thickness loss. But in this investigation, it was found that
this postulation does apply with the friction materials (Figure 7b). Thus, it could be
concluded that there is no direct correlation between hardness with average thickness wear
loss. Filip et al. (1995) reported that hardness of brake lining materials cannot be simply
related to the content of structural constituents, and there is no correlation between
hardness and wear resistance.
Figure 8 shows the relationship between the friction coefficient, average thickness loss and
porosity. Brake pad should have a certain amount of porosity to minimize the effect of water
and oil on the friction coefficient. Sasaki (1995) found that increasing porosity by more than
10 % could reduce the brake noise. It was observed that the sample with high porosity tend
to exhibit high friction coefficient. On the other hand, Figure 8b, indicates that there is no
direct correlation between average thickness loss with porosity.
During braking, the friction materials wear-off due to friction resistance between the friction
materials with the counter face material made of grey cast iron. Wear rate of friction depend
many factors such as operating parameters (temperature, speed, braking time), mode of
braking (continuous, intermittent braking), wear mechanism in operation during braking
(adhesion, abrasion, fatigue). When above the degradation temperature (230
C), the

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 15

Fig. 6. Normal and hot friction coefficient

Fig. 7. The relationship between physical properties of friction material used in this study:
(a) friction coefficient vs. hardness, (b) average thickness loss Vs. hardness.

Powder Metallurgy 16
binding properties of resin will become weak. As surface temperature increase with
increased braking times, the yield strength of the materials will be decreased and leads to
change in the wear mechanism and the real contact configuration as well as destruction of
friction film. Thus, the wear rate cannot be predicted based on physical and mechanical
properties because wear do not depend on material property but rather depends on
tribosystem property.

Fig. 8. The relationship between mechanical properties of friction material used in this
study: (a) friction coefficient vs. porosity, (b) average thickness loss Vs. porosity.
The friction and wear properties of friction materials depend on a number of different
factors such as pressure, speed, interface temperature, composition of friction material and
the metal member of the friction pair, duration and length of the friction path, friction
material density, its modulus of elasticity, type, design and geometry of friction mechanism
[Torovic, 1987]. From the analyses on test data, the following postulation could be made; (i)
higher hardness tend to reduce friction coefficient, (ii) higher porosity tend to increase
friction coefficient, (iii) there is no simple correlation between average thickness loss with
hardness and porosity. So in deciding with formulation can be used in the prototype
production, friction and wear test results are the main factor to be considered. Based on test
results, it could be concluded that all prototype samples complied with the requirements
and will be subjected to brake effective dynamometer tests.

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 17
3.2 Brake effective parameters: Friction coefficient and wear
Table 10 shows the friction and coefficient and wear results. Test results show that only
sample SM2 and SM9 have a minimum friction coefficient value below than 0.15 during the
first fade test sequence. Lower friction coefficient requires a longer braking distance before
the vehicle can be stop, which can cause road accident. In this test segment, the temperature
was increased from 100 to 550
C under the line pressure of 16 MPa. Under this condition,
the brake fade will take place. This fading effect is associated with the decomposition of the
organic binder which takes place between 250 and 475
C (Ramoussse et al., 2001). The friction
coefficient of the friction materials will vary with temperature and will fall off dramatically
as the contact temperature exceed the maximum organic decomposition temperature
depending on the ingredient and weight percentage used. However, in the second fade test,
the friction coefficient of the sample SM2 and SM9 have show a better result which is above
0.15, the minimum requirement. Characteristic friction after second fade also shows almost
recover to the characteristic friction in the early stage. Thus all the formulation developed
will be subjected to on-road performance test.
During braking process, brake pad is pressed against the brake rotor resulting in wear-off
the brake pad as well as the rotor material. Brake pad is designed as the sacrificial element
due to it low cost and ease of maintenance. Average values of wear detected after
completion of the brake effective dynamometer test procedure is given in Table 10. Wear
data are different for different formulation due to different ingredient and weight
percentage used in the composition. Wear characteristics is difficult to predict because it
depend on the physical, mechanical, chemical characteristic as well as the microstructure
changes during the braking process.

Average friction coefficient
Characteristic First fade
Thickness loss
COM 0.50 0.28 0.29 0.32 1.87
0.42 0.25 0.25 0.31 1.57
SM2 0.42 0.12 0.34 0.39 1.44
SM3 0.42 0.33 0.28 0.33 1.96
SM4 0.44 0.26 0.29 0.36 1.25
SM5 0.45 0.29 0.29 0.34 1.65
0.42 0.28 0.31 0.34 3.68
0.43 0.23 0.28 0.33 2.27
0.47 0.28 0.35 0.36 4.08
SM9 0.36 0.09 0.18 0.32 1.89
SM10 0.48 0.28 0.32 0.26 3.15
Table 10. Brake dynamometer test results
Figure 9 shows example of friction coefficient characteristics under different stops. The
friction coefficient characteristics for other samples vary for different stops as apparent from
Figure 9. The first and the second characteristic, first and second fade, and recovery
sequences reflect on the performance of brake lining. It can be seen from the fade test, the

Powder Metallurgy 18
coefficient of friction decreases with increased in temperature. This is attributed to physical
and mechanical, chemical and microstructural changes on the contact surface [Scieszka,
1980; Jacko, 1977; Talib et al., 2003; Ingo et al., 2004].

Fig. 9. Characteristic results of brake dynamometer tests
Even though the operating pressure, time and braking sequence of CHASE and
dynamometer is not the same, it supposed to produce the friction coefficient results of the
same trend for different composition. However, this postulation does not materialise in case
of friction materials. It can be seen from Figure 10 that only sample COM, SM1 and SM3
have higher friction coefficient when the samples were subjected to dynamometer tests as
compared to CHASE friction coefficient and the variation between friction coefficient
reading of CHASE and dynamometer is also not same. Thus, it could be concluded that
there is no direct correlation between the friction coefficient between CHASE and
dynamometer tests. This was taught due to dependent of friction coefficient with material
composition, microstructure and tribosytem.

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 19

Fig. 10. Bar chart of normal and characteristic friction of CHASE and dynamometer tests.
In case of hot friction, the variation between CHASE and dynamometer results is quite high
(Figure 11 ). This could due to the severity of test conditions applied during dynamometer
test. Surface temperature increases when the operating variables such as load, speed and
braking time are increased. In the dynamometer test, the surface temperature is increased
up to 550°C which much higher compared to CHASE test which is about 300°C. As the
surface temperature increases, the polymer materials will degrade. The onset of degradation
of the friction material starts at 230
C, and the degree of degradation increases with
temperature within the range of 269 – 400
C [Zhigao & Xiaofei; 1991]. The degradation of
the polymer materials may cause brake fade in which the friction is reduced as the
temperature increased [Rhee, 1971; Talib 2001]. The high temperature will also decrease the
yield strength and leads to changes in the wear mechanism and the real contact
configuration [So, 1996]. These phenomena could be the reason why the friction coefficients
during dynamometer test much lower than one during CHASE test. The different between
the friction coefficients for particular composition is not the same. This could be due to the
heterogeneous properties friction materials. Thus, it could be concluded that there no simple
correlation between the friction material under high temperature test condition when
subjected to CHASE and dynamometer tests.
Figure 12 shows the data of material thickness losses during braking tests on CHASE
machine and brake dynamometer. The results show there is no correlation between the two
test results. These variations can be due to the fact that CHASE machine uses a small
material sample (i.e. 1 inch x 1 inch x 0.25 inch) pressed against a large rotating drum that
does not represent the actual size of the lining material in its real intended application.
Whereas, the brake dynamometer evaluates a full size brake lining material as it is in real
application and thus simulating the actual braking condition of a vehicle. Thus, CHASE
machine is not recommended for evaluation of the thickness loss of the developed sample in
full size application. Ideally, all friction materials shall be tested and evaluated in all
conditions that they may encounter during their service such as under various brake
operating parameters (load, temperature and braking duration), road conditions (downhill
and winding roads) and wheather conditions (rain, sunshine and snow). For all these,

Powder Metallurgy 20
different vehicles will require different friction materials and unfortuantely, CHASE
machine perform rather poorly in predicting the actual performance of the materials when
they are put into real life application.

Fig. 11. Bar chart of thickness loss during CHASE and dynamometer test.

Fig. 12. Bar chart of thickness loss during CHASE and dynamometer test.
3.3 On-road performance
The development and validation of a friction material involve a significant amount of
testing in laboratory and on the road. As a vehicle is typically used under various road and
driving conditions, a friction material shall be tested in conditions closely representing these
driving conditions. Brake friction material developers will look for quantitative data from
these tests to evaluate their material formulations and track the effects of the modifications
that are made during the course of the product development. Of all the tests carried out

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 21
during a friction material development, on-road brake test is the final test normally
performed to evaluate and validate the formulation as the brake friction material is actually
tested under its real life application conditions. A brake test is basically a deceleration test
carried out between two speeds. Data taken during the test is used to calculate the time
taken, distance travelled and deceleration. A few other additional parameters such as brake
hydraulic pressure, brake pedal effort and temperature of the friction materials would also
be normally measured.
The developed formulations were subjected to on-road test as per ECE R13 and shall
achieve a minimum mean fully developed deceleration (MFDD) requirements as shown in
Table 11. Alternatively, the braking performance may also be evaluated in terms of the
stopping distance. The tests consists of three (3) test modes (cold, fade and recovery)
simulating real conditions of brake lining material temperature during its service. Figure 13
shows a typical display of Dewetron DEWE-5000 acquisition system during fade and
recovery tests.

Fig. 13. Display of data acquisition system; (a) fade test, (b) recovery test

Powder Metallurgy 22
Tests Mean fully developed deceleration
cold effective
heat fade

75 % of that prescribed and 60 % of figure recorded in the
cold effectiveness test
not less than 70 %, nor more 150 %, of figure recorded in the
cold effectiveness test
Pedal force Shall be more than 500N
Table 11. Minimum requirements of the performance tests (ECE R13)
Test results of on-road braking performance for all the developed friction material
formulation and a commercial pad are shown in Table 12 Figure 14. Out of the 10 prototype
and 1 commercial sample tested for on-road performance, four samples do not fully comply
with ECE’s requirements, namely, sample SM2, SM4, SM7 and SM9. Samples SM2, SM7 and
SM9 require pedal forces of 536N, 646N and 980 N, respectively under cold test conditions,
which are exceeding the maximum permitted pedal force of 500 N as shown in Table 11. As
such, further tests (i.e. fade and recovery) were not performed on these samples and the
samples were eliminated. Higher pedal force requires more driver effort to stop the vehicle,
which may stress the leg, especially for the lady driver. Sample SM4 does comply with
deceleration requirement under fade test with MFDD of 4.18 m/s
which is less than the
required value of 4.81 m/s
(i.e. >75% of 6.43). Test results also show that all other samples
(COM, SM1, SM3, SM5, SM6, SM8, SM10) fully comply with cold-, fade- and recovery-tests
requirements. Sample SM3, SM5, SM6 and SM8, though, shows higher values of MFDD
during recovery tests than the cold effectiveness tests, which is allowed by this regulation
which states that the MFDD can go up to 150% of the figure recorded in cold effective test.

Sample Test Mode
Pedal Force
Pad Temp. (deg C)
Left Right
Cold 138 7.65 93 126
Fade 150 5.49 350 456
Recovery 133 7.50 184 242
Cold 254 8.00 161 172
Fade 230 8.10 212 221
Recovery 218 8.24 158 140
Cold 536 8.14 169 177
Fade Not performed, F>500 N for Cold Test
Recovery Not performed, F>500 N for Cold Test
Cold 114 7.33 103 158
Fade 120 5.77 394 459
Recovery 116 7.39 202 260
Cold 96 6.83 115 110
Fade 94 4.18 416 312
Recovery 98 4.95 181 155
Cold 129 7.46 93 89
Fade 132 5.73 253 273

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 23
Sample Test Mode
Pedal Force
Pad Temp. (deg C)
Left Right
Recovery 136 7.67 139 137
Cold 128 8.1 155 154
Fade 130 7.07 272 260
Recovery 134 8.38 143 152
Cold 646 6.02 202 193
Fade Not performed, F>500 N for Cold Test
Recovery Not performed, F>500 N for Cold Test
Cold 133 7.29 111 99
Fade 135 5.67 345 289
Recovery 132 7.32 169 143
Cold 980 - 292 320
Fade Not performed, F>500 N for Cold Test
Recovery Not performed, F>500 N for Cold Test
Cold 104 7.05 316 277
Fade 109 5.72 515 527
Recovery 117 5.77 333 332
Table 12. On-road performance test results

Fig. 14. MFDD for Cold, Fade and Recovery Tests

Powder Metallurgy 24
In the braking process, kinetic energy of a moving vehicle is converted into thermal energy.
The generation of heat is due to the friction between the friction materials and brake disc.
The heat generated is dissipated to the surroundings by the brake disc and friction
materials. The ability of brake disc to dissipate this heat significantly affects the performance
and wear life of the friction materials. Heat generated during braking result a phenomenon
known as heat fade where the friction fall at elevated temperature. This fading effect is
associated with the decomposition of the organic compound. The decomposition of the
binder takes place between 250 and 475
C (Ramousse et al. 2001). This phenomenon results in a
reduction of friction as the temperature increased as observed by Rhee 1971 and Talib et al.
2001. This sudden drop of friction results in lower brake performance, in which longer
braking distance is required before the moving vehicle can be stopped.
Figure 15 shows that there is no direct correlation between thick loss of prototype brake pad
during brake dynamometer and on-road test. The test sequences and braking parameters of
the brake dynamometer and on-road tests are not the same. For homogeneous materials, the
thickness loss of the two test methods will be producing the same tend. But for friction
materials, this postulation does not apply. This was taught due to heterogeneous properties
of friction materials, where wear of friction materials is dependent on the mechanical,
chemical, thermal properties as well as the microstructure. On-road braking test results give
a better picture of the performance of the developed friction material formulations in real
life applications as compared with the laboratory test data.

Fig. 15. Thickness loss of brake pad during brake dynamometer and on-road test

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 25
3.4 Microstructural
Figure 16, shows the photograph of sample SM6 after subjected to on-road performance test.
The pad has pitting, grooving and moderate resin bleed. It was observed there is no flaking
or surface cracking evident on the worn surface of brake pad. The surface of rotor has light
lining transfer and light grooving.

Fig. 16. Post on-road performance test photograph on brake pad and brake disc
The worn surface after on-road performance test were analyzed using scanning electron
microscope equipped with energy-dispersive X-ray analyzer (EDX). Microstructureal
examaination on worn surface revealed that the mechanism composed a complex mixture of
abrasion, adhesion and delamination as shown in Figure 17. Figure 17a show a manifestation of
abrasion wear mechanisms where the harder peak asperities were ploughed into the surface.
Figure 17b and c show a manifestation of adhesion mechanism. Adhesion wear mechanism
composed a process of two-way transfer during sliding caused the formation of transfer layers
on both sides of the sliding surfaces (Figure 17b) as observed elsewhere (Kerridge & Lancaster
1956; Chen and Rigney 1985; Talib et al. 2003) due to mechanical alloying as reported by Chen
at el. (1984). Figure 17c shows transfer layers appeared to be sheared and flattened and smeared
on their surfaces during then raking process. Thus it can be concluded that the wear surfaces
became smoother with increase in braking time. Figure 17d shows a sympton of delamination
mechanism where it reveled the wear particles flake off from the wear surface when
reaching the critical length.

Powder Metallurgy 26

Fig. 17. Wear mechanism observed during braking process; (a) abrasion, (b) adhesion –
generation of transfer patches, (c) adhesion – smearing and (d) delamination
4. Conclusions
Characteristic of friction materials is very complex to predict and it is a critical factor in
brake system design and performance. To achive ideal brake friction material characteristic
such as constant a constant coefficient of friction under various operating conditions,
resistance to heat, water and oil fade, low wear rate, posses durability, heat stability, exhibits
low noise, and not to damage brake disc, some requirements have to be compromised in
order to achieve some other requirements. This can be done by changing the type and
weight percentage of the ingredients in the formulation. This works shows
The following phenomena could be postulated based on the physical and mechanical,
friction and wear test results using CHASE machine, brake effective test results using brake
inertia dynamometer, and braking performance test results;
i. Test results show that there is no simple relationship between the physical and
mechanical properties and thus, these test results could not be used to screen the
developed samples. The physical and mechanical properties are used for quality control
in producing friction material with consistent physical and mechanical properties.
Transfer patches

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 27
ii. Friction and wear characteristics obtained using CHASE machine could not be simply
related to physical and mechanical characteristics.
iii. Friction and wears assessment tests using CHASE brake lining machine can be used for
screening of friction material formulations during development as well as for quality
control. Thickness loss using CHASE machine cannot be used to predict thickness loss
using brake inertia dynamometer.
iv. The test sequence and parameters of brake dynamometer cannot simulate exactly all the
braking pameters and enviroment of on-road test condition. Thus there is no simple
correlation between the brake dynamometer test results with on-road performance
v. The final selection of the best formulation is based on on-road performance test results.
However, the prototype samples need to be subjected to endurance tests to ensure that
formulations can perform under real life application conditions.
The development and validation of a friction material involve a significant amount of
testing in laboratory and on the road. On-road brake test is the final test normally performed
to evaluate and validate the formulation as the brake friction material is actually tested
under its real life application conditions. Thus, vehicle testing on the test track is the
ultimate measure for the overall assessment of the brake performance testing and
evaluation. Out of the ten (10) developed friction formulations, only 6 samples complied
with the on-road braking performance requirements. However, further investigations on the
performance and wear of the developed brake pads need to be conducted on actual
intended application on various real road conditions.
During braking process, the brake pad is pressed against a rotating disc which in turn slows
down the rotation of the wheels of a vehicle and thus stops the vehicle. In the process of
decelerating a moving vehicle, kinetic energy is converted into thermal energy. This
accumulated heat is absorbed by the brake pads and brake disc before being dissipated to
the atmosphere. The accumulation of heat causes high surface temperatures in the lining
materials and the brake disc which leads to the changes to the mechanical, chemical and
wear mechanism. The brake lining materials wear off as a result of friction between the
lining materials and the brake disc. Micro-structural changes on the worn surface of the
brake reveals that the wear mechanisms operated during braking include adhesion,
abrasion and delamination. The wear mechanisms operated during braking are rather
complex with no single mechanism was found to be operating fully.
5. Acknowledgment
This research was supported by Ministry of Science, Technology and Innovation, Malaysia
by providing research grant and SIRIM Berhad for providing research facilities.
6. References
Begelinger, A. and Gee, A.W.J. 1973. A New Method for Testing Brake Lining Material.
ASTM Special technical publication 567, Philadelphia: American Standard for Testing
and Materials. Pp. 316-334.

Powder Metallurgy 28
Bros, J. & Sciesczka, S.F. (1977). The investigation of factors influencing dry friction in
brakes. Wear, Vol 34, pp. 13139.
Chen, L.H. & Rigney, D.A. 1985. Transfer during unlubricated sliding of selected metal
systems. Wear, Vol. 59, pp. 213-221
Cho, M. H.; Kim, S. J., Kim, D. & Jang, Ho. (2005), Effects of ingredients on tribological
characteristics of a brake lining: an experimental case study, Wear, Vol. 258, pp.
ECE Regulation R 13. Uniform Provisions Concerning the Approval of Passenger Cars with
Regard to Braking
Filip, P., Kovarik, L, Wright, M, A. 1995. Automotive brake lining chracterization. Proceeding of
the 8
international pacific conference on automobile engineering, 4-9 November,
pp. 417-422. Yokohama: Society of Automobile Engineers of Japan, pp. 34-422
Hsu, S.M.; Shen, M.C. & Ruff, A.W. (1997). Wear prediction for metals. Tribology International
30, pp. 377-383.
Ingo, G.M.: D’Uffizi, M., Falso, G., Bultrini, G., & Padeletti, G. (2004), Thermal and
microchemical investigation of automotive brake pad wear residues, Thermocimica
Acta, Vol. 418, pp. 61-68.
Jacko, M.G. (1977). Physical and chemical changes of organic disc pad. In: Wear of materials,
Glaeser, W.A., Ludema, K.C. & Rhee, S.K. (Ed). pp. 541-546. ASME, New York
Jacko, M.G.; Tang, P.H.S. & Rhee, S.K. (1984) . Automotive friction materials evaluation
during the past decade. Wear, Vol. 100, pp. 503 – 515
Jang, H.; Ko, K., Kim, S.J., Basch, R.H. & Fash, J.W. (2004). The effect of metal fibers on the
friction performance of automotive brake friction materials, Wear, Vol. 256, pp. 406–
JIS D 4418: 1996. Test Procedure of Porosity for Brake Linings and Pads of Automobiles
JIS D 4421: 1996. Method of Hardness Test for Brake Linings, Pads and Clutch facings of
Lim, S.C., Ashby, M.F. & Bruton, J.H. 1987. Wear-rate transitions and their relationship to
wear mechanisms. Acta Metall., Vol. 35, No. 6, pp. 1343 – 1348.
Lu, Y. (2006). A combinatorial approach for automotive friction materials: Effects of
ingredients on friction performance, Composites Science and Technology, Vol. 66, pp.
Mokhtar, M.O.A. 1982. The effect of hardness on the frictional behaviour of metals. Wear,
Vol. 78, pp. 297- 305.
More, A.J.W. & Tegart, W.J.McG. 1952. Relation between friction and hardness. Proc. Royal
Soc. A, Vol. 212, pp. 452-458.
MS 474: PART 1: 2003. Methods of Test for Automotive Friction materials (Brake Linings,
Disc Pads and Bonded Shoe): Part 1: Specific gravity (First Revision)
Mutlu, I.; Eldogan, O. & Findik, F. (2006), Tribological properties of some phenolic
composites suggested for automotive brakes, Tribology International, Vol. 39, pp.
Österle, W. & Urban, I. (2004), Friction layers and friction films on PMC brake pads, Wear,
Vol. 257, pp. 215-226.

Selection of Best Formulation for Semi-Metallic Brake Friction Materials Development 29
Rhee, S.K. (1973). Wear mechanism at low temperature for metal reinforced phenolic resins.
Wear, Vol. 2, pp. 261-263.
Rhee, S.K. (1976). High Temperature wear of asbestos reinforced friction materials. Wear,
Vol. 37, pp. 291-297.
Rhee, S.K. 1971. Wear of Material – Reinforced Phenolic Resins. Wear, Vol. 18, pp. 471-477.
Rigney, D. A. (1997). Comments on sliding of metals. Tribology international, Vol. 30, No. 5,
pp. 361-367.
SAE 661: FEB 97. Brake Lining Quality Test Procedure, Society of Automotive Engineers,
Warrendale, Pennsylvania.
SAE Recommended Practice J866 list for codes and associated friction coefficient , Society of
Automotive Engineers, Warrendale, Pennsylvania. Society Automotive of Enginner
SAE J2522: 2002. Dynamometer Global Brake Effectiveness Recommended Practice, Society
of Automotive Engineers, Warrendale, Pennsylvania
Sanders, P.G., Dalka, T.M., and Basch, R.H. (2001). A reduced-scale brake dynamometer for
friction characterization, Tribology International, Vol. 34, pp. 609-615.
Sasaki, Y. 1995. Development philosophy of friction materials for automobile disc brakes. The eight
international pacific conference on automobile engineering. 4-9 November, hlm.
407-412. Yokohama: Society of Automobile Engineers of Japan.
Scieszka, S.F. 1980. Tribological phenomena in steel-composite brake material friction pairs.
Wear,Vol. 64, pp. 367 – 378.
So, H. 1996. Characteristics of Wear Results Tested by Pin-on Disc at Moderate to High
Speeds. Tribology International, Vol. 29, No. 5, pp. 415 – 423
Talib, R.J. (2001) Investigation Into The Surface and Bulk Wear Morphology of Automotive
Friction Materials. PhD Thesis, National University of Malaysia, Bangi,
Talib, R. J.; Muchtar, A. & Azhari C.H. (2007). The Performance of Semi-Metallic Friction
Materials for Passenger Cars, Jurnal Teknologi, Vol. 47(A), pp. 53-72
Talib, R.J.; Muchtar, A. & Azhari, C.H. (2003). Microstructural characteristics on the surface
and subsurface of semi-metallic automotive friction materials during braking
process, Journal of Material Processing Technology, Vol. 140, pp. 694-699.
Talib, R.J.; Shaari, M.S., Ibrahim, W.M.A.W., Kemin ,S. & Kasiran, R. (2006). Properties
Enhancement of Indigenously Developed Brake Pad for Light Rail Transit, In: Brake
Friction Materials, Darius G. S. (Ed). Pp. 79-86, Shah Alam, ISBN 983-3644-72-4,
UPENA, Malaysia
Talib, R.J.; Azimah, M.A.B., Yuslina, J., Arif S.M. & Ramlan K. (2008), Analysis on the
Hardness Characteristics of Semi-metallic Friction Materials. Journal Solid State
Science & Technology, Vol. 16, No. 1, pp. 124-129
Tanaka, K.; Ueda, S. & Noguchi, N. (1973) . Fundamental studies on the brake friction of
resin-based friction materials. Wear, Vol. 23, pp. 349-365.
Todorovic, J. ( 1987) . Modelling of the tribological properties of friction materials used in
motor vehicles brakes. Proc. Instn. Mech. Engrs. Vol. C 226, pp. 911-916.
Tsang, P.H.S.; Jacko, M.G., and Rhee, S.K. (1985). Comparison of CHASE and Inertial
brake dynamometer testing of automotive friction materials, Wear, Vol. 103, pp.

Powder Metallurgy 30
Zhigao, X. and Xiaofei, L. 1991. A Research for the Friction and Wear Properties of a Metal-
fiber-reinforced Composite Material. In Mechanical Properties Materials Design.
International Symposia Proceedings 1991. Boqun Wu (ed.). Amsterdam: Elsevier
Science Publisher. pp. 611-615.
Porous Metals and Metal Foams
Made from Powders
Andrew Kennedy
Manufacturing Division, University of Nottingham, Nottingham,
1. Introduction
Porous materials are found in natural structures such as wood, bone, coral, cork and sponge
and are synonymous with strong and lightweight structures. It is not surprising that man-
made porous materials have followed and those made from polymers and ceramics have
been widely exploited. The commercialisation of porous metals has lagged somewhat
behind, but has received a boost following a surge in worldwide research and development
in the early 1990’s.
The unique combination of physical and mechanical properties offered by porous metals,
combinations that cannot be obtained with dense metals, or either dense or porous polymers
and ceramics, makes them attractive materials for exploitation. Interest mainly focuses on
exploiting their ability to be incorporated into strong, stiff lightweight structures,
particularly those involving Al foams as the “filling” in sandwich panels, their ability to
absorb energy, vibration and sound and their resilience at high temperature coupled with
good thermal conductivity.
The applications for porous metals (metals having a large volume of porosity, typically 75-
95%) and metal foams (metals with pores deliberately integrated into their structure through
a foaming process) depend on their structure. Closed cell foams, which have gas-filled pores
separated from each other by metal cell walls, have good strength and are mainly used for
structural applications. Open cell foams, which contain a continuous network of metallic
struts and the enclosed pores in each strut frame are connected (in most cases these
materials are actually porous or cellular metals), are weaker and are mainly used in
functional applications where the continuous nature of the porosity is exploited. Examples
of structures for both these types of porous metals are shown in Figure 1.
Table 1 summarises the potential uses for porous metals and metal foams, highlighting the
relevant attributes that make them suitable for that particular application (Ashby et al.,
2000). Specific examples of current applications for porous metals and metal foams can be
found in (Banhart, 2001), but it should be remarked that this is a dynamic area and new
applications and components are continually being developed.
This short review describes the main powder metallurgy-based manufacturing routes,
highlighting the key aspects of the relevant technology involved and the types of foam
structures that result.

Powder Metallurgy


Fig. 1. Micrographs of (left) a closed cell foam and (right) an open cell foam, or more
correctly, a cellular metal (Zhou, 2006).
Application Relevant Attributes
Lightweight structures Excellent stiffness-to-weight ratio when loaded in bending
Mechanical damping Damping capacity is larger than solid metals by up to 10x
Vibration control Foamed panels have higher natural flexural vibration
frequencies than solid sheet of the same mass per unit area
Acoustic absorption Open cell metal foams have sound-absorbing capacity
Energy absorbers /
Exceptional ability to absorb energy at almost constant
Heat exchangers Open-cell foams have large accessible surface area and high
cell-wall conduction giving exceptional heat transfer ability
Biocompatible inserts Cellular texture stimulates cell growth
Filters Open-cell foams for high-temperature gas and fluid filtration
Table 1. Potential application areas for porous metals and metal foams adapted from (Ashby
et al., 2000).
2. Processing methods for metal foams
There are many different ways to produce porous metals and metallic foams and these
methods are usually classified into four different types of production, using liquid metals,
powdered metals, metal vapour or metal ions. The use of powdered metals as the starting
material for foam production offers the same types of advantages (and often the same
limitations) as conventional powder metallurgical processes. If a particular metal or alloy
can be pressed and sintered, there is a high likelihood that it can be made into a porous
metal or metal foam.
2.1 Porous metals produced by powder sintering
2.1.1 Pressureless sintering
Loose pack, pressureless or gravity sintered metal powders were the first form of porous
metals and are still widely used as filters and as self-lubricating bearings. The porosity in
these components is simply derived from the incomplete space filling of powders poured

Porous Metals and Metal Foams Made from Powders

into and sintered in a die. With packing densities broadly in the range of 40-60%, but
affected by particle shape, size and vibration, the porosities in these structures are well
below those for most porous metals. The simplicity of the process means that porosity can
be included in a wide range of metals, limited only by the ability to sinter the metal in an
appropriate die. The process is most commonly used to sinter bronze powders to make
bearings, an example of which is shown in Figure 2, but porous structures from titanium,
superalloys and stainless steel have also been made in this way.

Fig. 2. Porous bronze made by pressureless sintering of approximately 100 m diameter
powders [Eisenmann, 1998].
In the case of Al alloys, where sintering is made difficult by the surface oxide layer covering
the powder particles, milling of the powder with sintering aids such as Sn and Mg is
required. The limitations imposed by the need to sinter in a die, principally on the size of the
product and the productivity, can be mitigated by performing die (or other) compaction
processes to increase the green strength of the compact so that it is sufficient to perform
containerless sintering. The inevitable consolidation that is involved will, of course, decrease
the already low porosity.
In an effort to increase the porosity in these parts, powders have been replaced by metal
fibres, made by processes such as melt spinning, which with higher aspect ratios, exhibit
lower packing densities, making them suitable for a wider range of applications (Anderson
and Stephani, 1999). A further development of this, leading to much higher porosities, is the
sintering of hollow metal spheres (which themselves are made via a powder route – to be
described later). Spheres with diameters ranging from 1.5 to 10 mm, with wall thicknesses
from 20 to 500 m can be arranged to produce both open and closed pore structures. Open
structures can be obtained in the same way as for powders and compaction can be used to
deform the spheres to polyhedral bodies, reducing the degree of open porosity. True closed
pore structures can be obtained by filling the interstices between the spheres with metal
powder followed by a sintering treatment. Porosity is contained both within and between
the hollow spheres and porosities in the range of 80-97% are reported (Anderson et al.,
2000). Examples of an open cell sintered hollow sphere structure and the spheres in cross
section, showing their thin walls, are presented in Figure 3.

Powder Metallurgy


Fig. 3. Steel hollow spheres (left) sintered to form an open cell foam and (right) sectioned.
The advantage of porous structures made from sintered powders or hollow spheres is that
there is good control of the volume fraction and to a lesser extent, the geometry and size of
the pores, leading to reproducible structures and properties. Sintered metal powder and
fibre structures are mainly suited to applications based on filtration, catalysis or heat
exchange. The low density structures that can be made from sintered hollow spheres can be
used for lightweight structural parts and for energy and sound absorption.
2.1.2 Gas entrapment
Internal porosity can also be developed in metal structures by a gas expansion (or foaming)
process based on hot isostatic pressing (HIPing) (Martin and Lederich, 1992). Initially,
following the standard method for HIPing of metal powders, a gas-tight metal can is filled
with powder and evacuated. In a deviation from common practise, the can is then filled
with argon gas at pressures between 3 and 5 bar before being sealed, isostatically pressed at
high temperature and then worked to form a shaped product, normally a sheet. Porosity is
generated by annealing the part. When holding at elevated temperature, the pressurised
argon gas present within small pores in the structure causes the material to expand (foam)
by creep. As HIPing is an effective method for sintering and the can material can be made
from the same material as the powder, this process could be used for many different metal
powders. The use of a can means that sandwich-type structures, consisting of a lightweight
foam core and thin, solid face sheets are produced. These types of structures are ideal for
lightweight construction.
Porous bodies with typically 20–40% of isolated porosity are obtained and theoretical
considerations show that no more than 50% porosity can be expected (Elzey and Wadley,
2001). Figure 4 shows a schematic representation of the process, which has been used to
make porous titanium sandwich structures for the aerospace industry, without the need for
complex joining methods. Disadvantages include low porosity and irregular-shaped pores.
2.1.3 Reactive processing
In contrast to the gas entrapment method, porosity is evolved much more rapidly when
foaming occurs in highly reactive multi-component powder systems such as those which
undergo self-propagating high temperature synthesis (SHS). The highly exothermic

Porous Metals and Metal Foams Made from Powders


Fig. 4. A schematic (left) of the processing steps used to manufacture titanium alloy foam
sandwich panels by gas entrapment (Ashby et al., 2000) and (right) the morphology of a
TiAl6V4 sandwich structure (Banhart, 2001).
reactions, initiated either by local or global heating of compacted powder mixtures to the
reaction ignition temperature, lead to vapourisation of hydrated oxides on the powder
surfaces and the release of gases dissolved in the powder. The reacting powder mixture
heats up rapidly to form a liquid containing (mostly hydrogen) gas bubbles and when the
reaction is complete, cools rapidly, entrapping the gas to form a foam. A schematic of this
process is shown in Figure 5 (Kanetake and Kobashi, 2006).
Gas formation and foam expansion can be augmented by the addition of vapour forming
phases such as carbon (which burns in air to produce CO) or foaming agents which react
together to increase the reaction temperature and produce fine particles that stabilise the
foam. As foaming takes place in the liquid state, stabilisation of the bubbles is needed to
avoid rapid collapse of the foam structure. Figure 5 shows how a reactive Ti+B
C foaming
agent increases the porosity in a Ni-Al powder mixture from 30 to 90% by 5% addition. It
can be seen that the pores are irregular in shape, as is the shape of the expanded foam.
Although the process is relatively simple, the production of foams is limited to combinations
of materials that react exothermically, some metal-metal systems but typically metals and
carbon or carbides, or metals and oxides. These limitations, coupled with the difficulty
controlling the expansion process and defining the shape of the expanded foam, mean that
few, if any, commercial foam products are made this way.

Powder Metallurgy


Fig. 5. A schematic (top) of the reactive powder process used to make metal foams and
(bottom) cross-sections of NiAl
foams containing different addition levels of a Ti + B
foaming agent mixture, after (Kanetake and Kobashi, 2006).
2.1.4 The addition of space-holding fillers
Many of the processes already mentioned have shortcomings in either the rather limited
levels of porosity that are achievable, or in the formation of pores that are highly irregular
with a wide distribution in sizes. A simple development of standard PM practices, but
incorporating a volume of sacrificial space fillers, offers a solution to both these problems.
Porous metals are produced by mixing and compacting metal powders with a space holder
which is later removed either during or after sintering, by dissolution or thermal
degradation, to leave porosity (Zhao and Sun, 2001). A schematic of this process is shown in
Figure 6. This simple method has the advantage that the morphologies of the pores and their
size are determined by the characteristics of the space holder particles and the foam porosity
can be easily controlled by varying the metal/space holder volume ratio. Addition levels of
the space holder are typically between and 50% and 85%. These are sufficiently high that
they are interconnected and hence can be removed easily. Above 85% the structure of the
struts is unlikely to be continuous and below 50%, residual space filler will be enclosed
within the structure, making removal very difficult.
Commonly, space fillers take the form of polymer granules and water soluble salts, but can
also be metal powders and ceramic or polymer hollow spheres (if the hollow spheres aren’t
removed then these materials are, strictly speaking, syntactic foams). Two distinct
approaches may be taken. In the first, the space filler can either thermally decompose,
sublime or evaporate below the sintering temperature of the metal matrix. This requires the
resulting skeletal metal structure to have good strength, which is affected during the
compaction stage. Polymer powders (for example PMMA), carbamide granules or Mg
grains are commonly used. In the second approach, the space filler has a higher melting
point than the sintering temperature and is removed, after sintering, by dissolution in a

Porous Metals and Metal Foams Made from Powders


Fig. 6. Production of an open cell foam by sintering a mixture of metal powder and a
removable agent (Ashby et al., 2000).
suitable solvent in which the metal matrix should be unaffected, most commonly water. In
the case of Al alloys, NaCl is the favoured space holder as it melts at approximately 800°C
and can be dissolved fairly rapidly in warm water. For higher melting point metals, NaAlO
with a melting point of 1800°C, is used.
In addition to the decomposition or melting temperature and general ease of processing,
including solubility in the solvent, space fillers are also selected based on their inertness and
lack of solubility in the matrix, as well as the availability in the size and shape desired
(typically in the range of hundreds of m to a few mm and often spherical space fillers are
preferred). As salts (such as NaCl) are brittle materials and, particularly when ground, are
angular in shape, melting and granulation methods have been used to produce them in
spherical form (Goodall and Mortensen, 2007). If these granules are porous, this has the
additional benefit of more rapid removal from the sintered part due to the granule being
able to disintegrate as well as dissolve. Figure 7 shows porous beads made by the controlled
agglomeration of fine salt particles and the structure of the resulting porous metal (Jinnapat
and Kennedy, 2010). The cell structure is clearly interconnected via windows between
neighbouring pores, the number and size of which are dependent upon the co-ordination
number and contact area for the packed spheres.
Processing difficulties can arise during mixing the metal powder with the often much larger
space holders. Segregation results in defects in the final product, usually in the form of
incomplete struts and cells. Liquid binders are used to promote homogeneous mixing, often by
ensuring that the space filler is coated with the metal powder. An alternative approach is to
vibrate the finer metal powders into the interstices within a packed bed of the larger space
holder particles. The metal powder – space holder mixtures are then compressed to form a
compact with sufficient strength to be sintered free-standing and, in the case of Al to disrupt
surface oxide films so that sintering can take place. Compaction processes can include all of
those used in standard metallurgical practices, including metal injection moulding.

Powder Metallurgy


Fig. 7. Images showing (left) porous spherical salt beads and (right) the microstructure of the
resulting cellular stainless steel (Jinnapat and Kennedy, 2010).
Despite a few drawbacks, for example if space fillers such as NaCl are not completely
removed this can lead to corrosion and the size of parts produced is rather limited, in part
due to slow dissolution of the space fillers which might take days even for small parts, this
method is a favoured route for the manufacture of porous metals from a wide range of
materials. It is particularly suited to those metals with high melting points and is a common
route for the production of porous biomedical devices made from Ti or NiTi.
2.1.5 Additive manufacturing
Porous metal structures can be built, layer upon layer, using processes such as selective laser
sintering (SLS), direct metal typing or 3D direct metal printing. 3D parts are constructed by
stacking these layers, the geometry of which is defined by a CAD model.
Direct metal laser sintering uses a high power laser to sinter metal powders on the surface of
a powder bed. After each cross-section is scanned, the powder bed is lowered by one layer
thickness, a new layer of material is applied on top, and the process is repeated until the
part is completed. Two-component powders can be used, comprising metals coated with a
polymer, where the laser only melts the coating. Products made in this way, however, still
need a secondary sintering step to produce sufficiently robust parts. Steel and titanium
foams can be made by both these methods, although with polymeric binders, contamination
of Ti with carbon occurs and direct metal laser sintering is, therefore, preferred.
Direct metal printing works in a similar way, spraying a polymeric binder, which is
dispensed through a print head, over a powder bed. 3D metal typing or screen printing uses
a metal powder mixed with a binder which is then spread over patterned masks and cured
layer by layer (Andersen et al., 2004). For both these processes, subsequent polymer-removal
and sintering steps are required. Figure 8 shows cellular structures made from 316L stainless
steel by direct typing. Among the benefits of these processes are high levels of design
flexibility, including small, precise cellular structures with complex internal geometries.

Porous Metals and Metal Foams Made from Powders


Fig. 8. Cellular structures made from 316L stainless steel by direct typing (Andersen et al.,
2.2 Metal powder slurry processing
2.2.1 Metal powders slurries
Slurry processing of ceramics has been used for many years to produce bulk products,
coatings, films and foams. The wealth of scientific research and resulting literature
pertaining to this area is not mirrored for metal powders slurries, the need for processing in
this way is perhaps not so great given the many alternative forming methods for metals. The
production of stable (non-agglomerated, non-sedimenting) slurries containing metal
powders provides significant challenges. Metal powders are larger in size than those used
for ceramic processing and have a higher density, making them more difficult to keep
suspended. Metal powder slurries, like their ceramic counterparts, are usually based on
aqueous systems to which a suspending agent (to increase the viscosity) and dispersants (to
prevent particle flocculation) are added (Kennedy and Lin, 2011). Other additions may be
needed to lower the surface tension, change the pH or facilitate gelation.
The advantages of using metal slurries as the basis for producing foams is that constraints
imposed by poor compressibility are removed, shaping can be performed by gelling in
simple rubber moulds and gas bubbles can be introduced into the system by whisking or
through the addition of gas-forming agents (although this means that the foam has to be
stabilised). Although not well established, there are likely to be limitations to the materials
that can be used based on interactions between the metal and the solvent (corrosion or
reaction) and the ability to form a stable slurry.
2.2.2 Slurry coating of polyurethane foams
Porous metals can be produced by a replication method using an open cell polyurethane
(PU) foam or sponge as a template (Quadbeck et al., 2007). In this process the polymer foam
is first coated with a metal powder slurry, usually performed by immersion or by spraying.
Excess slurry is removed by squeezing the foam, often by passing it through rollers. Without
this, the cells may become partially closed due to the formation of liquid films bridging the
cell struts. After coating and drying, the template is removed by thermal degradation and
the resulting, fragile metal skeleton is further heated to sinter the metal powder particles,
forming a rigid cellular metal structure.

Powder Metallurgy

Figure 9 shows a sintered porous stainless steel structure alongside the polymer foam used
in the replication process. The strut structure is also shown and it is clear from the cross
section that the struts are hollow (due to the space vacated by the polymer) and the walls of
the struts are porous. Open cell structures with cell sizes between 10–80 pores per inch
(about 2.5-0.2 mm) can be produced with precise pore structures with total porosities as
high as 96%. This total porosity also includes the porosity in the sintered metal struts, but
open cell porosities as high as 90% can be achieved.
This type of approach can be used to make metallic hollow spheres (which are themselves
sintered to produce cellular metals as was described earlier). To make hollow spheres,
expanded polystyrene spheres are coated with a metal powder slurry and then sintered,
during which the polystyrene is removed and the metal forms a dense metal shell
(Andersen et al., 2000).

Fig. 9. Images (left) showing a PU foam and a stainless steel foam made by slurry coating
and (right) the microstructure of the struts showing, inset, their hollow nature.
2.2.3 Slip reaction foam sintering
Metal powder slurries can also be foamed by the insitu generation of a gas. In the slip
reaction foam sintering (SRFS) process, the slurry (or slip) contains additives which stabilize
the slip during processing (Angel et al., 2004) and to this a solution of ortho-phosphoric acid
in either water or alcohol is then added. The hydrogen generated by the metal-acid reaction
creates bubbles in the slip causing it to foam. As the solvent evaporates during the drying
process, the pores formed by the hydrogen bubbles, which were originally closed, turn to
interconnected porosity and an open cell green part is obtained. Figure 10 shows the typical
structure for a steel foam which has irregular primary pore sizes as large as 3.5 mm and
secondary pores between 0.05-0.3 mm. After sintering the total porosity is typically 60%.
Foams have been produced from both steel and aluminium powders but for high porosities
the green strength is low and cracks form in the foamed material.
In a variation on this process the slurry is an aqueous polymer solution that has the ability
to form a gel (Shimizu and Matsuzaki, 2007). Gelation of the slurry is carried out by a
freezing and thawing process and after gelation, it is heated until the foaming agent
(hexane) decomposes, forming a gas, causing the gel to expand. The resulting foam is then
dried and sintered.

Porous Metals and Metal Foams Made from Powders


Fig. 10. Micrograph of a steel foam produced by the slip reaction foaming process (Angel et
al., 2004).
2.2.4 Foaming by mechanical whisking
Foams have been produced from ceramic slurries by introducing air into the slurry, in much
the same way as whisking cream (Sepulveda, 1997), and this method has recently been
translated to metal systems (Lin, 2011). In the same way as for ceramics, in order to stabilise
the air bubbles that are introduced during whisking, a surfactant must be added to the metal
powder slurry. Through sufficient aeration, a foam can be formed which can be poured into
a shaped mould made from almost any material. Despite the addition of surfactant,
drainage of the liquid from the network of pores does occur, inevitably leading to collapse
of the foam. To preserve the foam structure, slurry systems are designed to either gel by
heating (cellulose systems) or cooling (agarose systems) or be polymerised by the addition
of an initiator (acrylamide systems). The foamed body is then dried, further heated to burn
out the polymer and finally sintered to densify the matrix.
The pore structures for these foams are surprisingly uniform given the simplicity of the
process, showing round pores connected by small windows. Figure 11 shows a cross section
through a stainless steel foam in the gelled and dried condition, which demonstrates good

Fig. 11. A cross section (left) of a gelled and dried foam before sintering and (right) the foam
microstructure showing an open cell structure and (inset) the porous nature of the cell struts
(Lin, 2011).

Powder Metallurgy

green strength, and also presents the sintered microstructures for the cells and the porous
cell struts (inset). Porosities up to 90% have been achieved in sintered parts but work to date
has shown that it is difficult to vary the pore size beyond the range of 0.5-1.5 mm. It is
thought that this simple foaming process has the potential to produce foams that are
suitable for both structural and functional applications.
2.3 Foaming of compacted powder precursors
2.3.1 Foaming process
Metal foams can be produced by encapsulating a foaming (or blowing) agent into a
precursor made from compacted metal powder, followed by melting (Baumeister, 1990).
The foaming agents are fine powdered compounds that, when heated, decompose to form a
gas (typically they are metal hydrides or carbonates). When the compact is heated, usually
in a mould, above the solidus temperature of the alloy, which should also be above the
decomposition temperature of the foaming agent, the gas evolved causes expansion of the
precursor. Expansion is rapid but collapse occurs, requiring fast cooling to “freeze-in” the
foam structure. A schematic of the process used to make and foam powder precursors is
shown in Figure 12.

Fig. 12. The sequence of steps used to manufacture metal foams made from compacted
powder precursors (Ashby et al., 2000).
Figure 13 shows an expanded precursor that has been foamed in a metal mould, thereby
defining its cylindrical shape. Also shown is a radiograph of the sample, revealing the
porosity inside the part. The cross section of the foam, also shown in this figure, reveals that
although the pores are reasonably spherical, a large variation in the pore sizes is observed.
The closed porosity in foams made in this way is typically below 90%.

Porous Metals and Metal Foams Made from Powders


Fig. 13. Images of (left) a foam produced by the expansion of a melted PM precursor in a
mould, (centre) an X-ray radiograph of the foam structure and (right) a cross section of the
foam showing the pore structure.
2.3.2 Advantages and limitations of the process
The foaming of compacted powder precursors has several key advantages. It is one of the
few processes that can make closed cell foams, with the ability to produce near-net-shape
and complex foam parts, including foam plates and sandwich structures. Lightweight Al
foam parts made in this way are being used in a number of applications where their high
strength at low mass and excellent energy absorbing ability are exploited. The main
disadvantages are that large 3D parts are difficult to make and, in part due to the rapid
nature of the foam expansion process, the uniformity and reproducibility of the pore
structure can be unsatisfactory, leading to concerns over the reproducibility of the
mechanical performance (Kennedy, 2004).
Limits to the range of metals that can be foamed in this way arise due to several factors.
There is a requirement for the blowing agent to only start to decompose when the metal or
alloy is semi-solid. If decomposition occurs when the precursor is still solid, gas evolution
can cause cracking of the precursor and escape of the gas without it contributing to foam
expansion. It is, therefore, difficult to find suitable foaming agents for all metals. TiH
which starts to decompose at around 450°C (Kennedy

and Lopez, 2003), is used to foam low
melting point metals, such as Al and Mg, despite decomposing below the melting point of
most of their alloys. This requires the compacted precursor to contain very low (<2%)
porosity to contain the gas (Kennedy, 2002). This is not readily achieved by conventional PM
compaction methods and so to achieve the target densities, the use of high strength, pre-
alloyed powders is avoided, favouring mixed elemental powder additions, and powder
consolidation is either performed by hot die compaction, cold isostatic pressing followed by
extrusion or continuous powder rolling or extrusion processes. The foaming of reactive
metals and those with high melting points is not that practical, given the need for a
conductive metal mould to produce shaped parts, but SrCO
has been used to produce Fe-
based foams.
Foams produced by melting compacted powders are stabilised (at least temporarily) by
oxide films introduced into the liquid from the surface of the metal powders. The fraction of
3 mm

Powder Metallurgy

these oxides in the expanding liquid is critical to achieving good foam structures and stable
foams. Variations in oxide content for different powder sizes, from different suppliers or even
for different batches of powders (due to variations in the atomising or storage conditions) can
be the reason behind highly variable foaming responses (Asavavisithchai and Kennedy,
2006a). Varying the processing conditions during hot compaction can also affect the oxidation
of the metal powder and alter the foaming behaviour (Asavavisithchai and Kennedy, 2006a,
2006b). Figure 14 shows the effect of oxygen content on the foam expansion for an Al powder,
too low and foam collapse is severe, too high and the liquid is too viscous to foam. It should be
noted that the way in which this oxygen (or oxide) level is achieved, through heat treatments
at different temperatures or through atomisation, is not important.
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
heat treated at 500°C
heat treated at 550°C
as received



Oxygen content / wt.%

Fig. 14. The effect of oxygen content on the foaming of pure Al powder compacts, showing
that an optimum level is required (Asavavisithchai and Kennedy, 2006a).
3. Summary
An overview of some of the many and varied methods for making porous metals and metal
foams from metal powders has been presented. With research and development into metal
foams being vibrant and dynamic, pioneered by institutes like the Fraunhofer IFAM centres
in Bremen, Chemnitz and Dresden in Germany, the state-of-the-art is continually evolving
as the understanding behind powder processing, foaming and foam stabilisation improves.
For established foaming processes, research conducted within academia and industry has a
strong emphasis on eliminating problems which would otherwise limit the wider use of
these materials. This includes; improving the uniformity and reproducibility of the foam
structures, aiming to achieve uniform pore sizes and densities throughout the component
and similar foam structures from part to part; decreasing the processing and materials costs,
through improved or new processing routes, reduced waste and cheaper starting materials
and developing compelling case studies based on innovative design, simulation and testing
to demonstrate to end users that despite higher prices for some foams or components
containing foam elements, that these costs can be more that offset by the weight and energy
savings offered by these novel materials and structures.

Porous Metals and Metal Foams Made from Powders

4. References
Andersen O, Stephani G. 1999. Melt extracted fibres boost porous parts, Metal Powder Report,
54, 30-34.
Andersen O, Waag U, Schneider L, Stephani G, Kieback B. 2000. Novel metallic hollow
sphere structures, Advanced Engineering Materials, 2,192.
Andersen O, Studnitzky T, Bauer J. 2004. Direct typing: a new method for the production of
cellular P/M parts. In: Danninger H, Ratzi R, editors. Euro PM2004 Conference
Proceedings. Shrewsbury: European Powder Metallurgy Association , vol 4. p.189
Angel S, Bleck W, Scholz PF, Fend T. 2004. Influence of powder morphology and chemical
composition on metallic foams produced by Slip Reaction Foam Sintering (SRFS)-
process. Steel Res Intl, 75, 483-488.
Asavavisithchai S., Kennedy A.R., 2006. Effect of powder oxide content on the expansion
and stability of PM-route Al foams, Journal of Colloid and Interface Science, 297, 715-
Asavavisithchai S., Kennedy A.R., 2006. The Role of Oxidation During Compaction on the
Expansion and Stability of Al Foams Made Via a PM Route, Advanced Engineering
Materials, 8, 568-572.
Ashby MF, Evans A, Fleck NA, Gibson LJ, Hutchinson JW, Wadley HNG. 2000, Metal Foams:
A Design Guide. Butterworth – Heinemann, UK.
Banhart, J., 2001. Manufacture, characterisation and application of cellular metals and metal
foams. Progress in Materials Science,. 46(6), 559-632.
Baumeister J. 1990. German Patent 4,018,360,.
Eisenmann M. 1998. Metal powder technologies and applications. In: ASM Handbook, vol. 7.
Materials Park, USA: ASM International,1031.
Elzey DM, Wadley HNG. 2001. The Limits of Solid State Foaming, Acta Mater, 49, 849.
Goodall, R., Mortensen, A., 2007. Microcellular aluminium: Child's play. Advanced
Engineering Materials, 9, 951-954.
Jinnapat A, Kennedy A.R, 2010. The manufacture of spherical salt beads and their use as
dissolvable templates for the production of cellular solids via a powder metallurgy
route, Journal of Alloys and Compounds, 499, 43-47.
Kanetake N., Kobashi M., 2006. Innovative processing of porous and cellular materials by
chemical reaction, Scripta Materialia 54, 521–525
Kennedy A.R., 2002. The effect of compaction density on the foamability of Al-TiH
compacts, Powder Metallurgy, 45, 1, 75-79

A.R., Lopez V.H., 2003. The decomposition behavior of as-received and oxidized
foaming-agent powder, Mat Sci and Eng, A357, 1-2, 258-263.
Kennedy A.R., 2004, Aspects of the reproducibilty of mechanical properties in Al based
foams, J Mat Sci, 39, 3085-3088.
Kennedy, A R, Lin, X, 2011. Preparation and characterisation of metal powder slurries for
use as precursors for metal foams made by gel casting Powder Metallurgy, 54,3, 376-
Lin X, 2011. Foaming of stainless steel powder slurries, PhD thesis, University of
Nottingham, UK
Martin RL, Lederich RJ. 1992. Metal Powder Report, October, 30
Quadbeck P, Stephani G, Kuemmel K, Adler J, Standke G. 2007 Synthesis and properties of
open-celled metal foams. Mater Sci Forum, (534-536): 1005-1008.

Powder Metallurgy

Sepulveda P. 1997. Gelcasting foams for porous ceramics. Am Ceram Soc Bull, 76, 61-65.
Shimizu T, Matsuzaki K. 2007. Metal foam production process using hydro-gel and its
improvement. Mater Sci Forum, (539-543): 1845-1850.
Zhao, Y.Y., Sun D.X., 2001. A novel sintering-dissolution process for manufacturing Al
foams. Scripta Materialia,. 44, 1, 105-110.
Zhou, J., 2006. Advanced structural materials. Porous Metallic Materials, ed. e. Winston O.
Soboyejo., CRC Press , Taylor & Francis Group. USA, 22.
Aluminium Alloy Foams:
Production and Properties
Isabel Duarte and Mónica Oliveira
Centro de Tecnologia Mecânica e Automação,
Departamento de Engenharia Mecânica, Universidade de Aveiro,
1. Introduction
Ultra-light metal foams became an attractive research field both from the scientific and
industrial applications view points. Closed-cell metal foams, in particular aluminium alloy
(Al-alloy) ones can be used as lightweight, energy-absorption and damping structures in
different industrial sectors, detaining an enormous potential when transportation is
concerned. Despite the several manufacturing methods available, ultra-light metal foam
applications seem to be restricted to a rather less demanding market in what concerns final
product quality. The current available manufacturing processes enable effective control of
density through process parameters manipulation. However, none of them allow for
appropriate control of the cellular structures during its formation, leading to severe
drawbacks in what concerns final product structural and mechanical properties. The latter,
is undoubtedly the main reason for the lack of commercial acceptance of these ultra-light
metal foams in product quality highly demanding sectors, such as the automotive or
aeronautical sectors. The resolution of this problem is the main challenge of the scientific
community in this field. To accomplish the latter two approaches may be followed: i) to
develop new manufacturing processes or modify the existing ones to obtain foams with
more uniform cellular structures. (ii) to understand and quantify the thermo-physico-
chemical mechanisms involved during the foam formation in order to control the process,
avoiding the occurrence of such imperfections and structural defects.
Metal foams manufacturing processes seem to abound (Banhart, 2001) and can be classified
in two groups (Banhart, 2006): direct and indirect foaming methods. Direct foaming
methods start from a molten metal containing uniformly dispersed ceramic particles to
which gas bubbles are injected directly (Körner et al, 2005), or generated chemically by the
decomposition of a blowing agent (e.g. titanium hydride, calcium), or by precipitation of gas
dissolved in the melt by controlling temperature and pressure (Zeppelin, 2003). The indirect
foaming methods require the preparation of foamable precursors that are subsequently
foamed by heating. The foamable precursor consists of a dense compacted of powders
where the blowing agent particles are uniformly distributed into the metallic matrix.
Most commercially available metal foams are based on alloys containing (Degischer & Kirst,
2002): aluminium, nickel, magnesium, lead, copper, titanium, steel and even gold. Among

Powder Metallurgy 48
the metal foams, Al-alloys are commercially the most exploited ones due to their low
density, high ductility, high thermal conductivity and competitive cost. Some
manufacturing methods are already being commercialised. Direct foaming methods are
currently being commercial exploited in a large-scale, the following companies are just a few
examples. The Cymat Aluminium Corporation (in Canada) manufactures aluminium foams,
designated “stabilised aluminium foam” which is obtained by gas injected directly into a
molten metal (Degischer & Kirszt, 2002). Ceramic particles (e.g. silicon carbide, aluminium
oxide and magnesium oxide) are used to enhance the viscosity of the melt and to adjust its
foaming properties, since liquid metals cannot easily be foamed by the introduction of
bubbling air. The foamy mass is relatively stable owing to the presence of ceramic particles
in the melt. Foam panels with 1m in width and thickness range of 25-150 mm, can be
produced continuously without length limitations at production rates of 900 kg/hour. The
relative density of these foams is within the range 0.05-0.55 g/cm
. The average cell size is
2.5-30 mm. This process is the cheapest of all and allows for manufacturing large volume of
foams. Moreover, through this process it is possible to obtain low density foams. The main
disadvantage lies is the poor quality of the foams produced. The cell size is large and often
irregular, and the foams tend to have a marked density gradient. Despite this process
continuous improvement, the drawing of the foam and the size distribution of the pores are
still difficult to control. Besides its low production cost, secondary operations are usually
necessary. For example, the foamed material is cut into the required shape after foaming.
The machining of these foams can be problematic due the high content of ceramic particles
(10-30 vol. %) used in the process. Hütte Klein-Reichenbach Ges.m.b.H company (Austria)
also produces and commercialises aluminium foams with excellent cell size uniformity,
called MetComb. The process used is based on the gas injection method (Banhart, 2006) and
allows for the production of complex shaped parts by casting the formed foam into the
An alternative way for foaming melts directly is to add a blowing agent to the molten metal.
The blowing agent decomposes under the influence of heat and releases gas which then
propels the foaming process. Shinko Wire Company has been manufacturing foamed
aluminium under the registered trade name “Alporas” with production volumes reported
as up to 1000 kg of foam per day, using a batch casting process (Miyoshi et al, 2000). This
method starts with the addition of 1.5 wt.% calcium metal into the molten aluminium at
680°C, followed by several minutes stirring to adjust viscosity. An increase of viscosity is
achieved by the formation of calcium oxides. After the viscosity has reached the desired
value, titanium hydride (TiH
) is added (typically 1.6 wt.%), as a blowing agent by releasing
hydrogen (H
) gas in the hot viscous liquid. The melt starts to expand slowly and gradually
fills the foaming vessel. The foaming takes place at constant pressure. After cooling the
vessel below the melting point of the alloy, the liquid foam turns into a solid Al foam. After
that, the foam block is removed from the mould, it is sliced into flat plates of various
thicknesses according to its end use. This process is capable of producing large blocks of
good quality. Blocks with 450 mm in width, 2050 mm in length and 650 mm in height can be
produced. These foams have uniform pore structure and do not require the addition of
ceramic particles, which makes it brittle. However, the method is more expensive than
foaming melts by gas injection method requiring more complex processing equipment. The
density range of these foams is 0.18-0.24 g/cm
, and the mean cell size is about 4.5 mm.

Aluminium Alloy Foams: Production and Properties 49
Nowadays, foams manufactured by indirect foaming methods are also in the state of
commercial exploitation, but in small-scale by German and Austrian Companies, like
Schunk GmbH, Applied Light-weight Materials ALM and Austrian Company Alulight
GmbH (Banhart, 2006). Powder Metallurgical (PM) method is one of the commercially
exploited indirect methods to produce Al-alloy foams and it is also the research field of the
authors of this chapter. This process consists on the heating of a precursor material which is
obtained by hot compaction of a metal alloy (e.g. Al-alloy) with blowing agent powders (e.g.
), resulting in the foam itself. The metal expands, developing a highly internal porous
structure of closed-cells due to the simultaneous occurrence of the melting of the metal and
thermal decomposition of the blowing agent with the release of a gas (e.g. H
). The liquid
foam is then cooled in air, resulting in a solid foam with closed cells and with a very thin
dense skin that improves the mechanical properties of these materials. This process can
produce foams with porosities between 75% and 90%.
The PM method has several advantages in comparison with the methods described earlier
and will be further discussed ahead. The latter, has been addressed particularly in what
concerns two research lines: (i) the study of the physics and foaming technology, with
particular emphasis on dedicated process equipment development towards high quality
foams production. (ii) foam quality assessment through proper part property
characterisation, establishing its limits of application and seeking for new markets. This
chapter presents a detailed overview of the current state-of-art in what concerns to methods,
equipment and appropriate industrial procedures in order to obtain metal foams with good
quality. The advantages, the disadvantages and the limitations of this PM method are also
presented and discussed. An overview of the main challenges and perspectives in this field
concerning industrial implementation is also presented, whenever it is possible the authors
present novel research work.
2. Production of metal foams
Metal foam production by PM method can be divided into two production steps: (i)
production of foamable precursor and (ii) production of the metal foam itself through the
foaming of precursor material. A schematic diagram of the PM method is shown in Fig. 1.
The first step is the preparation of a dense solid semi finished product called foamable
precursor. The latter is attained by compacting a powder mixture containing the blowing
agent and the metal, by using a conventional technique. The second step includes the
production of the metal foam by heating this foamable precursor at temperatures above its
melting temperature.
This PM method can be used to produce foams of different metals and its alloys (Degischer
&Kriszt, 2002), such as aluminium and its alloys, tin, zinc, lead, steel and gold, which is one
of the advantage of this PM method. Among all metal foams, the Al-alloys are the ones that
have received more attention from both the research community and industry, due to its
enormous potential mainly in what concerns specific weight and highly corrosion
resistance. The most studied Al-alloys for foaming are pure aluminium, wrought alloys (e.g.
6xxx alloy series) or casting alloys (e.g. AlSi7Mg). The high quality foams of these different
metals can be obtained by choosing the appropriate blowing agent. Moreover, the
manufacturing parameters of the different stages require appropriate adjustment (Fig. 2).

Powder Metallurgy 50

Fig. 1. Powder Metallurgical metod for making metal foams.
Fig. 2. Manufacturing parameters of PM method (Duarte, 2005).
The main advantage of this PM method is to enable the production of components of metal
foams with different architectures (e.g. sandwich systems, filled profiles and 3D complex
shaped structures) in comparison with the others (Duarte et al, 2008, 2010). The materials
can be joined during the foaming step without using chemical adhesives (Duarte et al, 2006).
Other advantage lies in the fact that the addition of ceramic particles are not required,
avoiding the brittle mechanical behaviour inferred by these particles. Moreover, the foam
parts are covered by an external dense metal skin that improves its mechanical behaviour,
providing a good surface finish. The disadvantage of this PM process is the high production
cost mainly associated to the powder prices. Another disadvantage is the difficulty to
manufacture large volume foam parts. Nevertheless, sandwich panels of 2mx1mx1cm can
Cold isostatic pressing
Cold and hot pressing
Foamable tablets
Foamable rods
Foamable sheets
(tablets, rods
and sheets)
(into batch
3-D foam parts (any
geometries including
complex geometry)
Hollow structures
filled by foam
Sandwich panels
3-D foam parts with
metallic inserts
Bonding with
metal sheets
First production step: Production of foamable precursor
Second production step: Production of metal foam

Aluminium Alloy Foams: Production and Properties 51
already be manufactured (Degischer & Kirst, 2002). Furthermore, it must be pointed out that
during PM method it is still rather difficult to fully control the foaming process, which
results in lack of uniformity of the pore structure.
2.1 Preparation of the foamable precursors
The first step of the PM method to obtain metal foams is the production of foamable
precursor materials. The fundamental aspects of this production step are discussed ahead.
2.1.1 Selection of the powders
The blowing agent is a chemical compound which releases gas when heated, being the
responsible for the formation of bubbles. There are two main requirements to obtain high
quality foams. The first one is to ensure an uniform distribution of the blowing particles into
the metal matrix within the precursor material. The second is to ensure the coordination of
the thermal decomposition characteristics of the blowing agent and the alloy melting
behaviour to avoid cracks’ formation before melting. The selection of these powders is
therefore, detrimental for the foaming success. The characteristics of the metal powders, like
the purity, the particle size, the alloying chemical elements (content and type) and the
impurities (content and type), as well as the alloy melting behaviour and the thermal
decomposition characteristics of the blowing agent must be studied and known. The
literature highlights how powders from different manufactures could lead to notable
differences in foaming behaviour (Baumgärtner et al, 2000; Degischer & Kirst, 2002).
The blowing agents usually used for producing Al-alloy foams using the PM method are
metal hydrides, such as titanium hydride (TiH
), zirconium hydride (ZrH
) and magnesium
hydride (MgH
) (Duarte & Banhart, 2000). The employment of other blowing agent
powders, such as the carbonates, has been investigated as a cost-effective alternative to
metal hydrides (Haesche et al, 2010; Cambronero et al, 2009). It seems, though, that titanium
hydride still is the best choice for the foaming agent when producing Al-alloy foams, as
reported elsewhere (Duarte & Banhart, 2000). Fig. 3 presents a typical mixture of Al-alloy
and TiH
powders. The amount of the metal hydrides usually used is less than 1% in weight
in the initial powder mixture, based upon the Al or Al-alloy that is to be foamed. For
example, high quality of AlSi7 foams can be obtained by using 0.6wt.% in the initial powder
mixture (Duarte & Banhart, 2000).
The effect of the composition of the alloy, the impurities, the particle size and the alloying
chemical elements on the foaming behaviour has been studied (Duarte & Banhart, 2000;

(a) (b)
Fig. 3. Al-alloy and TiH
powders used in PM method.

Powder Metallurgy 52
Lehmus & Busse, 2004; Gokhale et al, 2007; Ibrahim et al, 2008; Helwig et al, 2011). Some of
these effects on the foam expansion behaviour or cell structure have not yet been well
established. A recent research of the effect of TiH
particle size on the foaming behaviour
and on the morphology of Al-alloy foam produced by PM process is reported (Ibrahim et al,
2008). These studies revealed that the use of the coarser particle sizes of TiH
leads to a
higher foam expansion and coarser macrostructure while the finer grade of TiH
leads to a
quite lower maximum expansion and a finer macrostructure. The use of different particle
sizes is an approach to adapt the onset of gas evolution temperature of the gas blowing
agent and improvement of the macrostructure of foamed aluminium.
The difference between thermal decomposition of the blowing agent and melting
temperature of the metal may cause the formation of irregular, crack-like pores in early
expansion stages, which then can lead to irregularities in the final foam (Duarte & Banhart,
2000). The research in this field has been demonstrating that the high-quality Al-alloy foam
is obtained when this difference is minimised. The basic rules to choose the best blowing
agent are related to the closing between the temperature of the beginning of the thermal
decomposition of the blowing agent and the liquidus temperature of the metal. This problem
has been approached in two different ways: (i) pre-treatments of the blowing agent powder
to delay the hydrogen release, i.e. to shift it to higher temperatures, (ii) to change the alloy
composition to obtain lower melting point through the addition of alloying elements.
Thermal pre-treatments that lead to partial decomposition and/or pre-oxidation of the
powder surface (Matijašević et al, 2006) or surface coatings (Proa-Flores & Drew, 2008) have
been applied. For example, TiH
powder particles are heated (480ºC) in air during a given
time (180 min) and an oxide layer is formed on the surface of the particles. This layer
delays gas release from the particles, so that hydrogen is ideally released during foaming
only after the alloy melting temperature has been reached. The powder colour depends on
the thickness of the formed oxide layer (Fig. 4). Another example, the TiH
powders can be
treated with acetic acid solution during 10 h, followed by a wash with distilled water until
the solution presented a low acidity. The particles can then be coated with a silicon dioxide
layer (Fang et al, 2005).

0 200 400 600 800 1000
air atmosphere
- untreated
- treated

Temperature (ºC)
- treated
- untreated


(a) (b) (c)

Fig. 4. Untreated (a) and treated (b) TiH
powders and their TG/DTG curves (c).

Aluminium Alloy Foams: Production and Properties 53
Another strategy is to change the alloy composition to obtain lower melting temperatures
through the addition of alloying elements, such as the magnesium, zinc and copper that
decreases the melting temperature (Helwig et al, 2011). Although this strategy appears to be
promising, research in this field has not been very systematic. The results revealed that these
treatments form a sufficiently thick oxide layer which leads to a minimum of hydrogen
loose. Helwig et al reported that the use of even higher magnesium amounts was found out
to lead to promising results in the run-up to this study. Researchers have been testing the
addition of the ceramic particles (e.g. alumina) in the initial powder mixture to improve the
foaming stability through the increase of the melt bulk viscosity (Kennedy, 2004). However,
the presence of these particles can originate a brittle mechanical behaviour of the foams.
2.1.2 Mixing of the powders
The mixing procedure should yield a homogeneous distribution of the alloying elements
and the blowing agent particles to ensure the high-quality of the Al-alloy foams with an
uniform pore size distribution. Blending the Al-alloy and the blowing agent powders is a
crucial step within the entire foaming process. This operation should be made to avoid the
agglomeration of the blowing agent particles and the alloying elements. This causes
structural defects and imperfections on the final foam. The mixers usually used are tumbling
mixers with or without alumina balls (Fig. 5). These balls do not add any other element to
the mixture, because the aluminium oxide is already present in the mixture.
An important practical aspect in the mixing operation is to obtain a clean and homogeneous
powder mixture. The impurities and solid powders by dirt, water or other particles
entrapped in the mixture may have a detrimental effect in foaming. These impurities can act
like nuclei uncontrolled voids during the thermal decomposition of the blowing agent
which will form larger pores at the latest foaming stages (Matijasevic & Banhart, 2006).

(a) (b)
Fig. 5. (a) Turbula mixer used to mix the powders. (b) Powder mixtures.
2.1.3 Compaction of the powders
The compaction should ensure that the blowing agent particles are embedded in the Al-
alloy matrix. The basic practical rule is to obtain a dense semi compact called foamable
precursor material with no residual open porosity in which theoretical density is close to
100% of the theoretical density of the aluminium matrix (Duarte&Banhart, 2000). The
production of precursors by compacting powder mixtures can be performed in a variety of
ways, e.g., by uni-axial (Kennedy, 2004), double-axial or isostatic pressing (Körner et al,
2000), extrusion (Baumgartner et al, 2000), powder rolling (Kitazono, 2004), etc., and all the

Powder Metallurgy 54
above mentioned techniques can be hot or cold. A conjugation of different conventional
compaction techniques can be used. Furthermore, the compacting process can be performed
in an inert atmosphere, in air or in vacuum (Jiménez et al, 2009). The most economical way
is the double-axial pressing in air, but the most efficient one is the high-temperature
extrusion. There are several procedures to produce foamable precursors. A simple process is
to compact the powder mixture using a hot uniaxial pressing at temperatures close to the
thermal decomposition of the blowing agent, using a pressing device and a die with a
heating system (Fig. 6). This process enables to yield more than 99% relative density of the
precursor (Fig. 7a).
Fig. 6. Die with the heating system used to prepare the precursor material.
The initial powder mixture is first compacted to cylindrical billets of 70-80% of the
theoretical density, using the cold isostatic pressing. Then, these billets are pre-heated to
temperatures close to the thermal decomposition of the blowing agent and extruded into
rectangular bars of various dimensions (Baumgärtner et al, 2000), as shown in Fig. 7b.
Kennedy reported that a minimum compaction density is required to achieve appreciable
expansion and is about 94% of the theoretical density (Kennedy, 2002). The highest value of
the expansion is obtained when the density is close to the theoretical density of the Al-alloy
matrix. The authors research results reveal that compaction of the mixture to achieve high
quality precursor materials must be performed at temperatures close to the initial thermal
decomposition temperature (e.g. 400ºC), in order to achieve the density value which leads to
a good foaming behaviour Moreover, the cold compaction should be done before the hot-
compaction in order to obtain a high level of densification (Fig.8).

(a) (b)
Fig. 7. Precursor materials of Al-alloy containing 0.6%wt. of TiH2 which is manufactured
using the laboratory system in Fig.6 (a) and supplied by IFAM (Baumgärtner et al, 2000).

Aluminium Alloy Foams: Production and Properties 55

(a) (b) (c)
Fig. 8. Precursor materials at different densification levels: (a,b) low-quality (c) high-quality.
Several parameters are evaluated to ensure the quality of the foamable precursors. The main
parameters are the density and the distribution of the blowing agent particles into the metal
matrix. The latter is usually evaluated by scanning electron microscope (Fig. 9).

Fig. 9. Foamable precursor sample of aluminium alloys containing 0.6 wt.% of TiH

composed with Al-matrix (dark gray) with Si (light gray) and TiH
(white colour) particles.
The required density of the foamable precursor is adjusted by manipulating the compaction
parameters (time and temperature). Duarte and Banhart reported that the hot-pressing
temperature is a very important parameter (Duarte & Banhart, 2000). The reduction of
compaction temperatures lead to insufficient compaction with some residual porosity (see
density values given in Fig.10 a). The hydrogen gas can escape from the melting alloy
without creating pores in this case. The same phenomenon is observed when the powders
are extruded instead of hot pressed. On the other hand when higher compaction
temperatures are employed lower maximum expansions can be observed, mainly because
some of the hydrogen is lost during compaction. Even higher compaction temperatures lead
to a rapid loss of foamability. The optimum compaction temperature therefore lies around
450°C for the case investigated, well above the initial decomposition temperature of TiH

(380°C). The compaction time is not considered a critical parameter for the compaction
temperatures chosen. The variations observed are not systematic and seem to be within the
range of normal statistical fluctuations.

Powder Metallurgy 56
150 200 250 300 350 400 450 500 550 600



hot pressing temperature (ºC)
00:00 00:10 00:20 00:30 00:40 00:50 01:00


Time [hh:mm]
500 °C
300 °C
200 °C



(a) (b)
Fig. 10. Effect of the hot pressing temperature on the precursor density (a) and on the
foaming behaviour (b).
2.2 Production of metal foams
The second step is the production of metal foams by heating the foamable precursor at
temperatures above the melting point of the alloy. The metal expands developing a highly
porous closed-cell internal structure due to the simultaneous melting of the aluminium and
thermal decomposition of the blowing agent (TiH
) in gas (H
). The solid foams are obtained
by controlled cooling of the formed liquid foam, at temperatures below the solidus
temperature. The foamed parts have a highly internal structure with closed cells (Fig. 11 b)
and are covered by a dense metal skin that improves their mechanical properties and
provide good surface finish (Fig. 11).
The foaming process usually takes place into the stainless steel closed moulds (Fig. 12). The
cavity of the mould should have the same design and dimensions of the final foam. The
foamable precursor into the mould is heated at temperatures above its melting point (Fig. 12
b). The material expands and fills the entire mould cavity. After that, the mould is cooled,
followed by the extraction of the foam (Fig. 12c). The furnaces used to manufacture the
metal foams usually are of the batch chamber furnace type (Fig. 13a). This research team has
developed a foaming continuous furnace to produce these materials (Fig. 13b).

(a) (b)
Fig. 11. Al-alloy foam covered by a dense skin (a), with a closed-cell internal structure (b).
This method enables the cost effective production of aluminium alloy parts without
limitations concerning shape ( e.g. panels, profiles or complex 3D shaped parts) (Duarte et
al, 2006, 2008, 2010). Here, the group research results are presented to illustrate some
practical examples (Fig. 14) which can be obtained by using the continuous foaming furnace

Aluminium Alloy Foams: Production and Properties 57
developed (Fig. 13 b). The disadvantage of the process is to produce components with a
large volume parts as it is the case of Shinko Wire method (Miyoshi et al, 2000).

(a) (b) (c)
Fig. 12. Practical aspects of the process. (a) closed mould. (b) mould with the precursor into
the pre-heated furnace. (c) Al-alloy foam block.

(a) (b)
Fig. 13. Foaming furnaces. Batch chamber furnace (a). Continuous foaming furnace
developed by this research team (b).

Fig. 14. 3D-parts of Al-alloy foams.
The atmosphere, heating rate and temperature of the thermal foaming cycle are some of the
manufacturing parameters which influence the quality and the properties of the resulting
foam. Moreover, it should also be referred that the characteristics of the mould (material,
design and dimensions), the type of the furnace (batch or continuous) should also be
considered (Duarte, 2005). The effects of these variables on the foaming behaviour may
therefore be investigated. To assess the latter, foaming tests are performed by heating the

Powder Metallurgy 58
precursor material up to its melting point inside of the apparatus called laser expandometer,
which was specially developed and constructed for this purpose (Fig. 15). Here, the
expansion (volume) and its temperature are monitored by means of a laser sensor and
thermocouple, respectively. The measurement of the volume of the expanding melt together
with the sample temperature generates a pair of functions V(t) and T(t) which characterise
the foaming kinetics. The foaming process is strongly governed by temperature effects
(Duarte & Banhart, 2000). The expansion curve is strongly dependent on the processing
conditions, mainly on the hot pressing temperature to obtain the foamable precursor
material, and the heating parameters during foaming (Fig. 10 b and Fig. 16).

Fig. 15. Laser expandometer used to characterise the foaming kinetics.

00:00 00:10 00:20 00:30 01:00 01:30 02:00 02:30

Time [hh:mm]

Fig. 16. Expansion curves of 6061 samples containing 0.6 wt. % TiH
, prepared with different
heating rates (Duarte & Banhart, 2000).

Aluminium Alloy Foams: Production and Properties 59
The final quality of the metal foams is evaluated by characterising its properties, namely the
density, structural and mechanical properties. The foam density is relatively predictable and
controlled by manipulation of the manufacturing parameters (Duarte & Banhart, 2005).
Higher heating rates lead to an earlier expansion of the foamable precursor because the
melting temperature is reached at an earlier time (Fig. 16). Besides that, the three expansion
curves for the highest heating rates are quite similar. Only significantly lower heating rates
lead to a change of the expansion characteristics, namely a lower maximum expansion. The
reason for this may be the gas losses due to diffusion of hydrogen and perhaps the strong
sample oxidation which might hinder expansion. In general, higher heating rate of the
foamable precursor leads to the formation of foams with lower density.
Although at an industrial stage, the process still has major limitations on the ability to obtain
tailor made cellular structure foams and to predict their properties. Cellular structured
foams with different sizes and shapes of pores, and structural defects, can present a high
density gradient depending on the component size (Fig. 17). These imperfections arise from
the difficulties in controlling the manufacturing process such as: (i) lack of homogeneity in
the precursor mixture (metal+blowing agent). (ii) lack of coordination between the
mechanisms of metal melting and thermal decomposition of the blowing agent.
(iii) difficulty in controlling the nucleation, growth and collapse stages. (iv) difficulty in
stabilizing the foam formed and preventing the cells’ collapse.

Fig. 17. Internal cellular structures for closed-cell AlSi7-alloy foam with different size.
3. The physics of metal foaming
The study of the physics of metal foaming is necessary to understand the underlying
principle of metal foam formation and stabilization in order to produce better foams. This
knowledge should lead to a control of these mechanisms during the foam formation.
Currently, there are no detailed studies on the quantification of these mechanisms, only
some aspects of foam evolution could be described by theoretical approaches. Despite the
longstanding interest and research efforts regarding these mechanisms, the simulation and
prediction of bubble nucleation in metallic foaming remains challenging. This is due to the
difficulties in observing these mechanisms under experimental or actual processing
conditions. For that, the researchers have been trying to quantify the involved mechanisms.
Several investigations on foam formation have been carried out. The experimental
techniques used for investigating the metallic foaming are mainly of two types: ex-situ
techniques (Babcsán et al, 2005) and in-situ techniques (Babcsán et al, 2007). In the ex-situ
techniques, the foaming process is interrupted by cooling at different foaming stages and

Powder Metallurgy 60
the resulting solid foam is characterized. The disadvantage of this approach is that it takes a
long time to carry out such investigation, and that the results suffer from a certain
inaccuracy originating statistical variations between a single experiment. Even if the starting
materials for the individual foaming tests were produced in the same way, each foaming
experimental test would turn out slightly differently, due to effects such as agglomerates of
the blowing agent particles, structural defects and impurities in the precursor (Duarte, 2005).
The 3D-image of X-ray tomographic observations of solid foam samples in different foaming
stages allow to observe the modification of the shape of the bubbles (Stanzick et al, 2002).
In the in-situ techniques, the foaming studies are evaluated during the evolution of one
single sample (Garcia-Moreno et al, 2005; Rack et al, 2009). Neutron radioscopy and X-ray
radioscopy were employed for in situ observation drainage mechanisms, the early stages
and growth stages, during the foaming process (Bellman et al, 2002). The temporal
development of the cellular structures of liquid metallic foams and the redistributions of the
metal can be observed by synchrotron based on neutron radioscopy.
Theoretical studies of the foaming process itself concentrate predominantly on the detailed
analysis of microscopic evolution of foams, mostly on the basis of an already existing
cellular structure (Stavans, 1993). A theoretical study on metal foam processing has treated
the material flow behaviour and focused on the description of the solidification stage during
the process using an one-dimensional model which combines the equations of foam
drainage with Fourier equation. However, this model is not able to describe the entire
foaming process starting from bubble nucleation to final foam development. Other authors
showed that the standard foam drainage equation (FDE) can principally be used to describe
drainage in metal foams (Körner et al, 2008; Brunke & Odenbach, 2006; Belkessam &
Fristching, 2003). However, the effective viscosity was considered to be one order of
magnitude higher than the original one to match the experimental observations (Stanzick et
al, 2002). A published model of metal foaming based on a Lattice–Boltzmann procedure
(Körner et al, 2002) treats the foaming problem in more detail, (i.e. from bubble nucleation to
the resulting foam structure, however, without considering the chemical decomposition of
the blowing agent as well as the thermal heating process). The foam is considered in the
liquid state; i.e, melting and solidification are not taken into account.
These studies have had significant roles in contributing to a more complete understanding
of bubble-growth phenomena. However, in studying bubble-growth behaviour, almost all
of these previous works involve pure theoretical studies without experimental verification,
since only limited experiments have addressed the dynamic behaviour of the phenomena.
Moreover, some of the physical parameters that were used to describe the materials adopted
in these theoretical studies were unrealistic. There are no mathematical models available for
foam evolution including nucleation, growth, coalescence and decay. Generally, numerical
or analytical models focus on a particular phenomenon (e.g. drainage). The analytical and
numerical approaches available in the literature are very limited. From the engineering
point of view it is a very complex process because it involves several physical, chemical,
thermal and mechanical phenomena that occur at the same time or successively.
3.1 Foam evolution
The evolution of the cellular pores during the foaming process has been evaluated and
discussed using the ex-situ and in-situ techniques, as shown in the presented examples in
Fig. 18 and 19, respectively.

Aluminium Alloy Foams: Production and Properties 61
The foaming process can be divided into three stages: bubble nucleation, bubble growth and
foam collapse (Fig. 18b). Duarte and Banhart have done some pioneer research, in the
understanding of the mechanisms involved in the PM process, which contributed to the
significant advance of the state of the art (Duarte & Banhart, 2000), using ex-situ technique in
which the foaming process is interrupted by cooling at different foaming stages and the
resulting solid foam is characterized. Topics such as bubble nucleation, bubble growth and
foam stability were discussed. As a result of this research, it is concluded that the bubbles
nucleation occurs usually in the solid state, the pressure generated by the gas can deform
the metallic matrix, the process is governed by the principle of semisolid metal processing,
the bubble growth dynamics is controlled by the thermal decomposition of the blowing
agent and the metal melting.

00:00 00:05 00:10 00:45 00:50 00:55 01:00


Time (hh:mm)



Fig. 18. (a) Expansion curve of precursor containing AA 6061 sample containing 0.6 wt. %
foamed in a pre-heated furnace at 800°C. (b) Morphology of the AA 6061 foams in
different foaming stages (foam diameter  30 mm). The letters show at which expansion
stage the sample was removed (see Fig. 15).
Foaming direction
Bubble nucleation Foam collapse Bubble growth
Coalescence effects
Drainage effects

Powder Metallurgy 62

Fig. 19. In-situ radioscopic images acquired during a foaming process of a precursor.
The foaming process can be divided into three stages: bubble nucleation, bubble growth and
foam collapse (Fig. 18). The shape of the bubbles varies during the foaming process. The
bubbles appear as cracks aligned perpendicular to the foaming direction, changing to a
spherical geometry, followed by polyhedral geometries. As described above, the PM
method consists on heating a solid precursor material (Fig. 9). When heated, the metal melts
(e.g. Al alloy) and the blowing agent (e.g. TiH
) produces a gas that creates the bubbles in
the foam. The heating process leads to partial metal melting as well as to the release of the
gas and consequently to the material foaming in its semi-solid state of the material. The heat
supply takes place in the solid material up to the solidus temperature. The decomposition of
the blowing agent may start in the still with the precursor in solid state, so that the bubble
nucleation can be previously initiated here. For this reason, the pores formed at early stages
of foaming appear as cracks aligned perpendicular to the foaming direction (Fig. 18b in the
nucleation stage). Moreover, the partial decomposition of the blowing agent particles during
the compaction step can occur in which the temperature in this step is near to the initial
decomposition temperature. The released gas in this compaction step is entrapping into the
matrix metal, and can create a sufficient number of initial nuclei into the foamable precursor
in which can act as centre of bubble nucleation. Gas accumulates in residual porosity and
builds up pressure as temperature increases. With the occurrence of bubble growth, a two-
phase flow (liquid metal and gas bubbles) develops above the solidus temperature.
After the alloy melting, the crack-like pores round off to minimize surface energy. Bubble
growth begins, driven by gas release from the blowing agent, and the structure starts to
appear as foam. With the increase of the temperature the internal gas pressure of each
nucleated bubble increase, and turn in the strength of the metal matrix is reduced down to
its value at the melting point. The bubble growth may not be uniform because depends on
the characteristics of the foamable precursor material. The elongated initial bubbles are
increasing in size and become more spherical (Fig. 18). The more spherical bubbles are
observed in the cellular structures when the expansion reaches the maximum value. The
spherical bubbles become more polyhedrical bubbles. After maximum expansion, no more
gas is released and the foam begins to collapse. The latter is due the drainage and
coalescence mechanisms, which are discussed ahead.
Foam growth depends not only on the rheology of the system metal/gas and mechanical
strength but also on the pressure inside each bubble and its architecture. This growth is
affected by various factors such as the content and distribution of the blowing agent, the
hydrostatic pressure or tension applied to the metallic matrix and the viscous properties of
Bubble nucleation Bubble growth

Aluminium Alloy Foams: Production and Properties 63
the system (metal+gas). The metallic foaming rheology is not simple and its mechanical,
thermal and chemical interaction leads to coupled problems of great complexity.
3.2 Foam: Collapse, stabilisation and solidification
The fundamental stability, collapse and solidification mechanisms during the foam
formation have been investigated. These topics are the most controversial ones in the metal
foam research. A better understanding of these mechanisms is required for an accurate
control of the foam structures, such as cell size and porosity. Ex-situ and in-situ techniques
have been used in this field. In-situ techniques have been used to measure the expansion, the
density evolution and the effects of the drainage and coalescence. The effects of the thermal
foaming cycle (heating rate, temperature and atmosphere) on the foam stability and foam
collapse parameters have been also studied and observed (Banhart, 2006). PM foams belong
to the class of transient (unstable) foams with lifetimes of seconds or to the permanent
(metastable) foams with lifetimes of hours. The foamability is thought to result from the
Gibbs–Marangoni effect where a membrane is stabilized during thinning due to liquid metal
flow towards the weakened region, because of a local increase in surface tension. The flow is
the response to a surface tension gradient. Due to viscous drag the flow can carry an
appreciable amount of underlying liquid along with it so that it restores the thickness. In
contrast to transient foams, the film thinning times in permanent foams are relatively short
compared to the lifetime. The stability is controlled by the balance of interfacial forces. These
forces equilibrate after drainage has been completed. The effect of temperature and gravity
have been evaluated via observation of bubble size, ruptures of the bubbles, relative
distribution of ruptures between bottom and top, draining and timing of draining, as well as
the effect of temperature distribution.
Foam decays by combination of three phenomena – gas loss, drainage and coalescence
(Duarte, 2005). Part of the gas is lost by diffusion from the outer surface to the surrounding.
In the liquid state, loss is expected to rise due to a higher diffusivity than in its solid state.
However, crack formation during expansion at the outer surface of precursors can also lead
to gas losses. Moreover, bubble rupture at outer surface also results in sudden gas loss. The
second and third way of gas loss are random events and not much effort was attributed to
asses their contribution except for some qualitative statement. Drainage is one of the driving
forces for the temporal instability of liquid foams, caused mainly by gravitational and
capillary effects. The drainage and coalescence mechanisms can be observed using ex-situ
techniques with the formation of a thick metal layer on the bottom of the solid foam and
large pores, respectively. Foam shows a very complex rheological behaviour including the
bubble deformation, rearrangement and avalanches processes.
Foam solidification is also an essential processing step of foam production. An uncontrolled
solidification can create defects (e.g. cracks) in the cell wall of the foams (Duarte, 2005). So,
this study helps to reduce the defects observed in final structures and thereby to improve
the mechanical performance of the foam. The solidification by cooling of a foamed liquid
metal is a “race against time", in as much as the relatively heavy liquid is prone to drainage,
which rapidly reduces the foam density and hence provokes instability and collapse. The
foamed liquid is immediately subjected to gravity-driven drainage of liquid, creating a
vertical profile of density (or liquid fraction). At any point in the sample this adjustment
must proceed until the freezing point is reached. Thus, at intermediate times, the sample

Powder Metallurgy 64
consists of a solidified outer shell surrounding a draining liquid core. The competition
between drainage and heat transfer, leading to solidification was studied by Mukherjee et al
using X-ray radioscopy (Mukherjee et al, 2009). A hitherto unknown expansion stage was
observed during solidification of Al-alloy foams. The phenomena that occur simultaneously
while foam solidifies are associated to its volume change (Mukherjee et al, 2009). The extra
expansion is observed and it is an anomalous behaviour since the foam is expected to
contract during solidification. This extra expansion takes place whenever the combined
volume gain rate is more than the combined volume loss rate. Moreover, it increases as the
cooling rate decreases. Mukherjee reported that the slow cooling of foams can trigger extra
expansion which in turn can induce defects in foam morphology. This is due to the extra
expansion induced rupture during solidification of the metallic melt inside the cell wall.
4. Properties
The properties of the metal foams belong to a group of materials called cellular solids which
are defined as having porosity >up to 0.7 (Gibson & Ashby, 1997). Natural foams are
produced by plants and animals such as cork or bone. Man made foams can be
manufactured from a variety of materials such as ceramics, polymers and metals. There are
two categories of foams: open - and closed- cells. Here, it is presented a brief overview of the
main properties of closed-cell Al-alloy foams obtained by the PM method.
Metal foams combine properties of cellular materials with those of metals. For this reason,
metal foams are advantageous for lightweight constructions due to their high strength-to-
weight ratio, in combination with structural and functional properties like crash energy
absorption, sound and heat management (Asbhy et al, 2000; Degischer & Kriszt, 2002).
Many metals and their alloys can be foamed. Among the metal foams, the Al-alloy ones are
commercially the most exploited due to their low density, high ductility, high thermal
conductivity, and metal competitive cost.
4.1 Structural properties
There are several structural parameters of these foams, such as number, size-pore
distribution, average size, shape and geometry of the pores, thickness, intersections and
defects in the cell-walls and thickness, defects and cracks of the external dense surface for
describing the cellular architecture of the foams. The properties of these foams are
influenced by these morphological features (Gibson &Ashby, 1997; Ramamurty & Paul,
2004; Campana et al, 2008).
Progress has been made in understanding the relationship between properties and
morphology. Although this exact interrelationship is not yet sufficiently known, one usually
assumes that the properties are improved when all the individual cells of a foam have similar
size and a spherical shape. This has not really been verified experimentally. There is no doubt
that the density of a metal foam and the matrix alloy properties influence the modulus and
strength of the foam. All studies indicate that the real properties are inferior than the
theoretically expected due to structural defects. This demands a better pore control and
reduction in structural defects. Density variation and imperfections yield a large scatter of
measured properties, which is detrimental for the metal foams reliability (Ramamurty & Paul,
2004). Wiggled or missing cell-walls reduce strength, and in turn, result in a reduced

Aluminium Alloy Foams: Production and Properties 65
deformation energy absorbed under compression (Markaki & Clyne, 2001; Campana & Pilone,
2008). Mechanical studies demonstrate that selective deformation of the weakest region of the
foam structure leads to crush-band formation (Duarte et al, 2009). Cell morphology and
interconnection could also affect thermal and acoustic properties (Kolluri et al, 2008). It is
widely accepted that foams with a uniform pore distribution and defects free, are desirable.
This would make the properties more predictable. Only then, metal foams will be considered
reliable materials for engineering purposes and will be able to compete with classical
materials. Despite their quality improvement in the last 10 years the resulting metal foams still
suffer from non-uniformities. Scientists aim to produce more regular structures with fewer
defects in a more reproducible way which is the crucial challenge of the research in this field.
Foam characterisation results revealed that the cellular structures of the Al-alloy foams
obtained by PM method have pores with different sizes and shapes (Fig. 20). A large size
distribution of the cellular pores with irregular cell shape is observed. The closed pores are-
mostly- of polyhedral or spherical geometry (Fig. 20 and 21). Spherical pores with a thick
thickness of cell-wall are mostly observed in the bottom and lateral sides of the foam
samples (Figs. 20 and 21a). Polyhedral pores with a thin cell-wall thickness are mainly
observed at the top of the foam samples (Figs. 20 and 21b). The distribution of the solid
metal in the foam is also non-uniform and leads to a higher density gradient (Fig.20). These
materials have a broader cell diameter distribution curve (Fig. 20c). The cell-size distribution
is dominated by high number of the small pores. The most of the pores have diameter lower
than 2mm. The magnified images of the cross section of the sample reveal small porosities in
the dense surface skin (Fig. 22). Significant morphological defects such as cracks or spherical
micropores in cell walls and cell wall wiggles and dense surface skin are also observed. Each
cell has normally approximately 5 other ones in its vicinity (Fig. 21). The distribution of the
cell-wall thickness has an asymmetric shape for these foams (Fig.21). The smallest cell-wall
thickness is about 70 m. The maximum cell-wall thickness is about 500 m. The thickness
of the cell wall depends on the foam density. The thickness of the external dense surface
skin around sample varies, where the higher values are located in the lateral and bottom
sides. AlSi7 foams presents 565.56 m, 365.40 m and 214.58 m, respectively for bottom,
lateral and top sides of the samples (Fig. 22). Other structural feature that affects the
mechanical behaviour is the microstructure of the massive cell material. Depending on the

<2 mm 2-4 mm 4-6mm >6mm
Pore range [mm]


(a) (b) (c)
Fig. 20. Cellular structures of AlSi7 (a) and AlSi1Mg (b) foams. Cell-pore size distribution as
a function of the number of cell pores for both foams (c).

Powder Metallurgy 66
alloy composition and on the manufacturing process, metallic dendrites, eutectic cells,
precipitates, or even particles can be observed in the final cellular structures. Foams having
same density but made of different Al-alloys can reached different plateau stress.

(a) (b) (c)
Fig. 21. Pore geometries: (a) spherical, (b) polyhedral and (c) other geometries.

(a) (b) (c)
Fig. 22. Thicknesses of the dense surface skin in different sections a sample AlSi7 foam: (a)
top, (b) lateral and (b) bottom sides.
4.2 Mechanical properties
Many literature studies have been undertaken on the mechanical properties of metal foams.
A broad survey of the understanding of the mechanical behaviour of a wide range of
cellular solids is provided by Gibson and Ashby (Gibson & Ashby, 2000). Others, have
carried out experiments to investigate the behaviour of metallic foams under different
loading conditions, particularly the properties of metal foams under impact loading. The
possibility of controlling the load-displacement behaviour by an appropriate selection of
matrix material, cellular geometry and relative density makes foams an ideal material for
energy absorbing structures. Among the several mechanical testing methods available,
uniaxial compressive mechanical tests are commonly used to evaluate the compressive
behaviour and the energy absorbed of these foams. The elastic modulus, yield and plateau
strengths are the most important mechanical properties parameters which are obtained from
these curves. The stress-strain curves of closed-cell Al-alloy foams display either plastic or
brittle fracture depending on foam fabrication and microstructure (Sugimura et al, 1997;
Banhart & Baumeister, 1998).
The compression behaviour of these Al-alloy foams depends on several parameters such as:
(i) the Al-alloy composition; (ii) the foam morphology (cell size range); (iii) the density
gradient of samples; (iv) the defects of cellular structure (cell walls) and (v) the
characteristics of the external surface skin. The influence of the density and the architecture

Aluminium Alloy Foams: Production and Properties 67
of these foams on the mechanical properties are strong and complex. Depending on the
material from which the foam is made, different mechanisms (brittle or ductile) can be
observed. The compressive properties, such as, the average plateau stress, modulus, the
elasticity and the energy absorption, depend, above all, on the foam density in which their
values increase with the rise of the density (Fig. 23).
0 10 20 30 40 50 60 70 80 90 100
0 10 20 30 40 50 60 70 80 90 100
AlSi7 foams
0.47 g/cm
0.54 g/cm
0.55 g/cm
0.58 g/cm
0.58 g/cm


Strain (%)
High density High density

0.38 g/cm
0.39 g/cm
0.42 g/cm
0.44 g/cm
0.51 g/cm
0.52 g/cm
0.55 g/cm
0.56 g/cm
Strain (%)
AlSi1Mg foams

Fig. 23. Compressive stress-strain curves for the different specimens (cylindrical samples
height/diameter ratio equal 1: h== 30mm x 30mm).
As it can be depicted from Fig. 23, the stress-strain curves are divided into three
characteristic regions. The first region (I) is linear-elastic where the load increases with
increasing compression displacement almost linearly (elastic deflection of the pore walls),
followed by a plastic collapse plateau (Region II) with a nearly steady compression load
(pore walls yield or fracture, whereas increasing deformation does not require an increase of
the load). The last region is the densification of the foam (Region III) where there is a rapid
increase on the load after the cell walls crushed together.
These foams exhibit, after an elastic loading, a more or less clear plateau region. This plateau
stress is important to characterise the energy absorbing behaviour and is a good material
property for the compression performance of a foam. The measurements of the plateau
stress depending on the different methods exist for measurement of the plateau stress
depending on the course of the stress-strain curve.
The failure modes and mechanisms associated to these foams at different regions of the
load-displacement have been identified (Fig. 24).
The elastic deformation occurs due to bending of the edges, elongation of cell walls and
trapped gas pressure inside the cells. The deformation is not visible (point B, in Fig.24b). The

Powder Metallurgy 68
latter is almost totally reversible, and occurs uniformly throughout the sample. After
reaching the elastic limit, the collapse of the cells starts, mostly by distortion (stretching),
rotation and/or sliding of the edges and cell walls, with permanent deformation (points C in
Fig. 24b). A progressive collapse of the cells was observed in the “plateau” stage of the load-
displacement curve (points C to F, in Fig. 24b). This deformation is not uniform due to the
irregular structure of the foam (pore size distribution, thickness of cell walls, etc.) (Fig. 20a).
The slope that characterises this region may be related to the compression of fluid trapped
inside the cells, or due to tensile stress in the cell walls. The slope increases with increasing
density of the foam. The shape of collapsed cells is very different from its original shape, as
it contains bended and distorted cell walls that may even touch each other. However, in
general, the fracture of cell walls does not occur. The initial collapse begins in a small group
of cells in the region with the lowest local density of the sample. Collapse does not occur in
all cells (points E, in Fig.24b), starting in the cells that are less resistant or with higher loads.
The collapse of a cell induces the collapse of neighbouring cells. Moreover, the collapse of
neighbouring cells evolves in successive layers and eventually leads to the formation of a
single deformation band (points E in Fig.24b).
Al-alloy foams are often used as filler material in lightweight structures subject to crash
and/or high velocity impact or as thermal/acoustic insulation devices. The energy


Fig. 24. Load-displacement curve of a AlSi7 foam under compressive loading. Cellular
structures of deformed AlSi7 foam samples at different cross-head displacements.
0 2 4 6 8 10 12 14 16 18 20 22 24 26

Cross-head velocity: 1.0 mm/min
Temperature: 22ºC
Samples number: 4
AlSi7 Foam

Displacement [mm]
B: 1mm C: 2mm D: 4 mm E: 8 mm F: 16mm

Aluminium Alloy Foams: Production and Properties 69
absorption capability of these foams can be well estimated from the stress-strain
compression behaviour of the material which is estimated from the area under the stress-
strain curve (Fig. 25a). As foam materials exhibit a constant stress “plateau” they can absorb
higher levels of energy than dense aluminium alloys. Most of the absorbed energy is
irreversibly converted into a plastic deformation energy which is a further advantage of
foamed Aluminium. For the same stress level, the dense material is deformed in the regime
of reversible linear-elastic stresses, releasing most of the stored energy when the load is
removed. Al-alloy foams exhibit higher energy absorption capabilities (Fig. 25). The increase
of the energy absorption with increasing foam density is clearly obvious (Fig.25b).

0 20 40 60 80 100


Strain (%)
0 20 40 60 80 100
AlSi7 foams
0.47 g/cm
0.54 g/cm

0.55 g/cm
0.58 g/cm
0.58 g/cm
0.58 g/cm

0.60 g/cm

Strain (%)

(a) (b)
Fig. 25. (a) Absorbed energy per volume in a certain strain interval is the area under the
stress-strain curve. (b) Absorbed energy curves of AlSi7 foam under compressive loading at
different densities.
5. Challenges
Despite its technological advances, the metallic foam formation is not problem free and still
poses challenges. Questions related to a very hot topic (i.e. the control of the pores size and
shape of metal foams) that is seen with alacrity by the scientific community due to potential
applications of these materials in the transport industry, are highlighted and discussed in
detail. The key-question is how to produce metal foams, in series, achieving uniform cellular
structure, in order to improve the manufacture reproducibility and to control foam
architecture. A key goal of this group research work is to develop the missing knowledge to
fill in the highlighted gap in the production of Al-alloy foams of uniform closed-cell
structures and transfer it to industry.
6. References
Ashby, M.F., Evans, A., Fleck, N.A., Gibson, L.J., Hutchinson, J.W. & Wadley, H.N.G. (2000).
Metal foams — a design guide, Butterworth-Heinemann, ISBN 0-7506-7219-6, London,
Babcsán, N., García-Moreno, F. & Banhart, J. (2005). Metal foams—high temperature
colloids: Part I. Ex situ analysis of metal foams. 261(1-3):123-130, 2005.

Powder Metallurgy 70
Babcsán, N., García-Moreno, F. & Banhart, J. (2007). Metal foams - High temperature
colloids. Part II: In-situ analysis of metal foams. Colloids and Surfaces A:
Physicochemical and Engineering Aspects, Vol. 309, No. (1-3), pp. 254–263, ISSN 0927-
Banhart, J. (2001). Manufacture, characterisation and application ofcellular metals and metal
foams Progress in Materials Science Vol. 46, No. (6), pp. 559–632, ISSN 0079-6425
Banhart, J. (2006), Metal foams: production and stability, Advanced Engineering Materials, Vol.
8, No. (9), pp. 781–794, ISSN 1438-1656
Banhart, J. (2008). Gold and gold alloy foams. Gold Bulletin, Vol. 41, No. (3), pp. 251-256,
ISSN 0017-1557
Banhart, J. & Baumeister, J. (1998). Deformation characteristics of metal foams. Journal of
Materials Science, Vol. 33, pp. 1431–1440, ISSN: 0022-2461
Baumgärtner, F., Duarte, I. & Banhart, J. (2000). Industrialisation of P/M foaming process.
Advanced Engineering Materials, Vol.2, No.4, pp. 168-174, ISSN 1438-1656
Belkessam, O. & Fritsching U. (2003). Modelling and simulation of continuous metal foaming
process. Modelling and Simulation in Materials Science and Engineering, Vol. 11, No.6,
pp. 823–837, ISSN 0965-03932003.
Brunke, O. & Odenbach, S. (2006). In situ observation and numerical calculations of the
evolution of metallic foams. Journal of Physics: Condensed Matter, Vol. 18, pp. 6493–
6506, ISSN 0953-8984
Campana, F. & Pilone, D. (2008). Effect of wall microstructure and morphometric
parameters on the crush behaviour of Al-alloy foams. Materials Science &
Engineering A, Vol. 479, No. (1-2), pp. 58-64, ISSN 0921-5093
Cambronero, L.E.G., Ruiz-Roman, J.M., Corpas, F.A. & Ruiz Prieto, J.M. (2009).
Manufacturing of Al–Mg–Si alloy foam using calcium carbonate as foaming agent.
Journal of Materials Processing Technology, Vol. 209, pp. 1803–1809, ISSN 0924-0136
Degischer, H.P. & Kriszt, B. (Ed(s).) (2002). Handbook of Cellular Metals, Wiley-VCH, ISBN 3-
527-30339-1, Weinheim
Duarte, I. & Banhart, J. (2000). A study of aluminium foam formation – kinetics and
microstructure. Acta Materialia, Vol.48, No.9, pp. 2349-2362, ISSN 1359-6454
Duarte, I., Banhart, J. Ferreira & M. Santos. (2006) Foaming around fastening elements.
Materials Science Forum, Vol. 514-516, pp. 712-717, ISSN 0255-5476
Duarte, I., Santos, M. & Vide, M. (2008). Processo contínuo de produção de peças e
protótipos em espumas metálicas. Ingenium, II série, Março/Abril 2008, Vol. 104,
pp. 78-80, ISSN 0870-5968I.
Duarte, I. (2005), Espumas Metálicas: Processos de fabrico, Caracterização e Simulação
numérica. PhD thesis. FEUP- Faculty of Engineering of University of Porto, Porto.
Duarte, I., Teixeira-Dias, F., Graça, A. & Ferreira, AJM. (2010). Failure Modes and Influence
of the Quasi-static Deformation Rate on the Mechanical Behavior of Sandwich
Panels with Aluminum Foam Cores. Mechanics of Advanced Materials and Structures,
Vol.17, No. (5), pp. 35-342, ISSN 1537-6494
Fang, J., Yang., Z., Zhang, H. &Ding, B. (2005). The coating process of silica film on TiH2
particles and gas release characteristic. Chemical Engineering Science, Vol. 60, pp.
845-850, ISSN 0009-2509

Aluminium Alloy Foams: Production and Properties 71
Garcia-Moreno, F. , Babcsan, N. & Banhart, J. (2005). X-ray radioscopy of liquid metal foams:
influence of heating profile, atmosphere and pressure. Colloids and Surfaces A:
Physicochemical and Engineering Aspects, Vol. 263, pp. 290–294, ISSN 0927-7757
Gibson, L.J. & Ashby, M.F. (1997). Cellular solids – Structure and properties, Second Edition,
Cambridge University Press, ISBN 0-521-49911-9, Cambridge, United Kindgom
Gokhale, A., Sahu, S.N., Kulkarni, V.K.W.R., Sudhakar, B., Rao & N. Ramachandra (2007)
Effect of Titanium Hydride Powder Characteristics and Aluminium Alloy
Composition on Foaming. High Temperature Materials and Processes. Vol. 26, No.4,
pp. 247–256, ISSN 0334-6455
Haesche, M., Lehmhus,D., Weise, J., Wichmann, M. & Mocellin, I. C.M. (2010). Carbonates
as Foaming Agent in Chip-based Aluminium Foam Precursor. Journal of Materials
Science & Technology, Vol. 26, No. (9), pp. 845-850, ISSN 1005-0302
Helwig, H.-M., Garcia-Moreno, F. & Banhart, J. (2011). A study of Mg and Cu additions on
the foaming behaviour of Al–Si alloys. Journal of Materials Science, Vol. 46, No. (15),
pp. 5227-5236, ISSN 0022-2461.
Ibrahim, A., Körner, C. & Singer, R.F. (2008). The effect of TiH2 particle size on the
morphology of Al-foam produced by PM process. Advanced Engineering Materials,
Vol. 10, pp. 845–848, ISSN 1438-1656
Lehmhus, D. & Busse, M. (2004). Potential New Matrix Alloys for Production of PM
Aluminium Foams. Advanced Engineering Materials, Vol. 6, pp. 391-396, ISSN 1438-
Jiménez, C., García-Moreno, F., Mukherjee, M., Görke, O., Banhart, J. (2009). Improvement
of aluminium foaming by powder consolidation under vacuum. Scripta Materialia,
Vol.61, No.5, pp., 552–555, ISSN 1359-6462
Markaki, A.E. & Clyne, T.W. (2001). The effect of cell wall microstructure on the
deformation and fracture of aluminum-based foams, Acta Materialia, Vol. 49, N.o 9,
pp. 1677-1686, ISSN 1359-6454
Matijasevic, B. & Banhart, J. (2006). Improvement of aluminium foam technology by
tailoring of blowing agent. Scripta Materialia, Vol. 54, No. (4), pp. 503-508, ISSN
Matijašević, B., Banhart, J., Fiechter, S., Görke, O., Wanderka, N. (2006), Modification of
titanium hydride for improved aluminium foam manufacture, Acta Materialia, Vol.
54, N.o 7,pp. 1887–1900, ISSN 1359-6454
Miyoshi, T., Itoh, M., A. kiyama, S. & Kitahara, A. (2000). Alporas aluminum foam:
production process, properties, and applications. Advanced Engineering Materials,
Vol. 2, No. 4, pp. 179-183, ISSN 1438-1656
Mukherjee, M (2009). Evolution of metal foams during solidification, Technischen Universität
Proa-Flores, P.M. & Drew, RAL, (2008).Production of Aluminum Foams with Ni-coated
TiH2 Powder. Advanced Engineering Materials, Vol. 10, No. (9), pp. 830-834, ISSN
Kennedy, A. R. (2002). Effect of compaction density on foamability of Al-TiH2 powder
compacts. Powder Metallurgy, Vol. 45, No. (1), pp. 75-79, ISSN 0032-5899
Kennedy, A. R. (2004). Effect of foaming configuration on expansion. Journal of Materials
Science. Vol. 39, pp. 1143-1145, ISSN 0022-2461

Powder Metallurgy 72
Kitazono, K., Sato, E. & Kuribayashi, K. (2004), Novel manufacturing process of closed-cell
aluminum foam by accumulative roll-bonding. Scripta Materialia, Vol. 50, No. (4),
pp. 495–498, ISSN 1359-6462
Kolluri, M., Mukherjee, M., Garcia-Moreno, F., Banhart, J. & Ramamurty, U. (2008). Fatigue
of a laterally constrained closed cell aluminum foam. Acta Materialia. Vol. 56, pp.
1114, ISSN 1359-6454
Körner, C., Thies, M. & Singer, R. F. (2002). Modeling of Metal Foaming with Lattice
Boltzmann Automata. Advanced Engineering Materials, Vol. 4, N.o 10, pp. 765 - 769,
ISSN 1438-1656
Körner, C., Arnold, & Singer, R. F. (2005). Metal foam stabilization by oxide network
particles. Materials Science and Engineering A, Vol. 396, N.o (1-2), pp. 28–40, ISSN
Körner, C. (2008). Foam formation mechanisms in particle suspensions applied to metal
foams. Materials Science and Engineering A, Vol. 495, N.o (1-2), pp. 227-235, ISSN
Rack, A., Helwig, H.-M., Bütow, A., Rueda, A., Matijasevic´-Lux, B., Helfen, L., Goebbels, J.
& Banhart, J. (2009). Early pore formation in aluminium foams studied by
synchrotron-based microtomography and 3-D image analysis. Acta Materialia
57(16): 4809–4821, 2009, ISSN 1359-6454
Ramamurty, U. & Paul, A. (2004). Variability in mechanical properties of a metal foam, Acta
Materialia, Vol. 52, No. (4), pp. 869-876, ISSN 1359-6454
Sugimura, Y., Meyer, J., He, M.Y., Bart-Smith, H., Grenstedt, J. & Evans, A.G. (1997). On the
mechanical performance of closed cell Al alloy foams. Acta Materialia, Vol. 45, No.
(12), pp.5245-5259, ISSN 1359-6454
Stanzick, H., Helfen, L., Danikin, S. & Banhart, J. (2002). Material flow in metal foams
studied by neutron radioscopy. Applied Physics A: Materials Science & Processing,
Vol. 74, N.o (1), pp. 1118-1120, ISSN 0947-8396
Stavans, J. (1993). The evolution of cellular structures. Reports on Progress in Physics,Vol. 56,
pp. 733-89, ISSN 0034-4885
Zeppelin, F., Hirscher, M., Stanzick, H. & Banhart, J (2003). Desorption of hydrogen from
blowing agents used for foaming metals. Composite Science and Technology, Vol. 63,
pp. 2293–2300, ISSN: 0266-3538
The Fabrication of Porous Barium Titanate
Ceramics via Pore-Forming Agents (PFAs)
for Thermistor and Sensor Applications
Burcu Ertuğ
Gedik University,
Department Of Metallurgical & Materials Engineering, Yakacık/Kartal, Istanbul
1. Introduction
The perovskite family includes many titanates used in various electroceramic applications,
for example, electronic, electro-optical, and electromechanical applications of ceramics.
Barium titanate, perovskite structure, is a common ferroelectric material with a high
dielectric constant, widely utilized to manufacture electronic components such as mutilayer
capacitors (MLCs), PTC thermistors, piezoelectric transducers, and a variety of electro-optic
devices (Wang, 2002).
For positive temperature coefficient of resistance (PTCR) and humidity/gas sensing
applications of BaTiO
, a porous microstructure is required. When oxygen is adsorbed on
the grain boundaries, PTCR effect is enhanced. Thus a porous structure is needed for PTCR
properties. Besides, porous ceramics can also be used for humidity/gas sensing. The water
vapour is adsorbed on pores improving the electrical conductivity of the surface.
A number of routes are employed in order to fabricate porous ceramics for PTCR and/or
humidity/gas sensors. Low pressure forming can be used for the production of BaTiO

ceramics. Forming can be done by unaxial pressing since hot pressing or hot isostatic
pressing can eliminate pores. Also several other pressureless forming techniques can be
used to make porous bodies. Another method for fabricating porous ceramics is using pore
forming agents (PFAs) prior to sintering. PFAs form porosity through the ceramic body by
different mechanisms. Different porosifiers that can be used for the production of barium
titanate ceramics are starch based, carbon based, metallic or polymeric porosifiers. In the
following paragraphs, crystal structure of, donor doping of and the electrical properties of
barium titanate is described. Also the concept of porosity is briefly mentioned. Porous
ceramics production, pore formers in general and pore formers in BaTiO
is also explained.
Finally PTCR and humidity/gas sensing properties of BaTiO
is described in detail.
2. Perovskites and barium titanate (BaTiO
oxides which adopt the perovskite structure form an important group of compounds
possessing many useful and interesting physical properties. The idealized structure is cubic,

Powder Metallurgy

point group m3m. The structure can be thought of in two ways. First, it can be viewed as
being constructed from corner-sharing octahedral of oxygen ions, with the small B cations
occupying the octahedral site and the much larger A cations occupying the interstices
between the stacked octahedra. Alternatively, the structure can be thought of as close-
packed ordered layers of oxygen and the larger A cations (AO
), with the B cations
occupying the octahedral interstices between the layers (Brook, 1991). The perovskite lattice
of barium titanate is shown in Fig.2.1.

Fig. 2.1 The crystal structures of barium titanate (BaTiO
). (b) [001], (c) [110], and (d) [111]
directions (Hutagalung, 2011).
exhibits two basic structures, one being a ferroelectric, tetragonal, perovskite
structure within temperature range 5-125°C. Above 125°C, the BaTiO
has the cubic
perovskite, paraelectric phase (a=4.009A°, c/a=1). BaTiO3 exhibits a first-order displacive
phase transition at 125°C from cubic (m3m) to tetragonal (4mm) on cooling. This change is
accompanied by an elongation in the [001] direction and a contraction along the a-axes.
Below 5°C, BaTiO
has another displacive phase transition resulting in an orthorhombic
(mm2) structure (a=5.667A°, b=5.681A°, c=3.989A°). Associated with the phase change,
there is a continuous change in the polarization direction, from parallel to the <100>
direction of the unit cell, towards the <110> direction. At -90°C, BaTiO
is transformed into
the rhombohedral crystal structure and the polarization direction becomes parallel to the
<111> direction (Zhou, 2000). The change of the crystal lattice dimensions with temperature
in BaTiO
is shown in Fig.2.2.
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications


Fig. 2.2 Lattice constants of BaTiO
as function of temperature (Wang, 2002).
In an ideal cubic perovskite, the ionic radii, r
(i=A, B, O), satisfy the relation r
+ r
= √2 (r

+ r
). The Goldschmidt tolerance factor for perovskites is, hence defined by
t = {r
+ r
} / {√2 (r
+ r
) } (2-1)
For pure barium titanate t~1.06. t is greater than 1 due to the fact that the Ti
ion is smaller
than its cavity (BaZrO
has t~1) and/or that the Ba
is larger than its cavity (SrTiO
has t~1)
(Tsur et al., 2001).
2.1 Donor doping of BaTiO

In a stoichiometric solid solution the extra positive charge of a donor center can be
compensated by a cation vacancy or an anion interstitial
(2MO) = 2D˙
+ Oo +O’’
(3MO) = 2D˙
+ V‘’
+3Oo (2-3)
Eq. (2-2) maintains a perfect cation sublattice, while the anion sublattice is perfect in Eq.(2-3).
A shift from the stoichiometric solid solution with its compensation by a lattice defect, to a
non-stoichiometric solid solution with compensation by an electronic defect, requires an
interaction with the ambient atmosphere. The stoichiometric solution may gain or lose
oxygen at oxygen activities greater than or less than, in equilibrium with the stoichiometric
composition. In this way, The compensating lattice defect is eliminated and is replaced by
electrons in the case of donor dopants (Smyth, 2000).
When La
replaces Ba
on the A-site (La is too large to replace Ti on the B-site), charge
imbalance is created which must be compensated by either cation vacancies on the A- or B-
site (ionic compensation), or by electrons (electronic compensation) as follows
+ 3TiO
= 4La˙
+ 3Ti
+ 12Oo
’’’’ (2-4)

Powder Metallurgy

+ 3TiO
= 2La˙
+ 3Ti
+ 9Oo
’’ (2-5)
+ 2TiO
= 2La˙
+ 2Ti
+ 6Oo
+ 2e’ (2-6)
Ionic compensation ((2-4) and (2-5)) should have negligible effect on the room temperature
conductivity due to the immobility of cation vacancies; La doped BaTiO
compensated in
this way should, therefore, remain insulating. In contrast, electronic compensation (Eq. (2-6))
should cause a substantial increase in conductivity, in which the number of carriers equals
the La concentration (Morrison et al., 2001).
Fig.2.3 indicates the effect of donor concentration (La
) on the electrical conductivity and
grain size of BaTiO
. As the donor concentration increases initially the conductivity
increases up to 0.15% of La
and then decreases up to 0.3% of La
. The high conductivity
region is where the electronic compensation dominates, after a critical donor concentration,
cation vacancy compensation dominates and electrical conductivity decreases. Up to 0.3% of
, the grain size of BaTiO
is not affected by donor concentration being 25 μm, however,
above 0.3% of La
, grain size decreases to 5μm.

Fig. 2.3 Schematic of donor concentration influence on room temperature electrical
conductivity and grain size.
Donor dopant incorporation is achieved by either electronic compensation at low
concentrations or vacancy compensation at high concentrations. High concentrations of
segregating donors at grain boundaries inhibites grain growth. At small concentrations,
donor incorporation by electronic compensation explains the high conductivity. As the
average dopant concentration increases, the local donor concentration at the grain boundary
increases rapidly due to segregation. The donor incorporation at the grain boundary shifts
from electronic to vacancy compensation, resulting in the formation of highly resistive
layers and also, grain size decreases due to significant dopant drag on the boundary
mobility (Desu, 1990).
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

3. Positive temperature coefficient of resistance (PTCR) property
Barium titanate is a very attractive material in the field of electroceramics and
microelectronics due to its good characteristics. Its high dielectric constant and low loss
characteristics make barium titanate an excellent choice for many applications, such as
capacitors, multilayer capacitors (MLCs) and energy storage devices. Doped barium titanate
has found wide application in semiconductors, positive temperature coefficient resistors,
ultrasonic transducers, piezoelectric devices, and has become one of the most important
ferroelectric ceramics (Vijatovic et al., 2008).
Barium titanate (BaTiO
) is an insulator with a large energy gap of 3.05 eV at room
temperature. It can be made n-type semiconducting by partially substituting Ba
trivalent cations such as La
and Y
or Ti
by pentavalent cations such as Nb
and Sb
The extra positive charge is compensated by free electrons. Doped BaTiO
which is sintered
in air, exhibits a very high relative permittivity and a sharp increase in resistivity above
Curie temperature, i.e 120°C (called “positive temperature coefficient of resistance”, or
PTCR) when the structure changes from ferroelectric to paraelectric. Pure barium titanate is
an insulator with no PTCR effect. A high PTCR effect can be achieved by the incorporation
of transition metals such as Mn, V, Fe, Cu and Cr. This is because the transition metals
substitute for Ti
as a lower valency state and act as acceptors (Kim, 2002; Kim et al., 2000;
Park, 2004).

Fig. 3.1 Electrical resistivity for typical PTCR device and schematic presentation of defect
chemistry responsible for PTCR effect (Vijatovic et al., 2008).
The grain-boundary model given by Heywang treats the grain boundary as a n-type Shottky
barrier with deep acceptor states at grain boundaries (Vijatovic et al., 2008). The resistance
anomaly behavior of doped BaTiO3

is shown in Fig.3.1.
The double-Schottky energy barrier model proposed by Heywang is one of the most
effective model which explains the mechanism of PTCR effect. According to this model, the
double-Schottky type energy barrier is formed along the grain boundary and its barrier
height determines the total conductivity. The electrical potential barrier results from

Powder Metallurgy

segregated acceptor ions or adsorbed oxygen at the grain boundaries. Sintering under
vacuum or reducing gas atmosphere strongly reduces PTCR effect (Zhang, 2004).
Heywang explained PTCR effect in terms of potential barriers. The double Schottky barriers
form at the grain boundaries. These barriers result from electron trapping by acceptor states
at the interfaces. The acceptor states arise from segregated impurities at the grain
boundaries or from adsorbed oxygen or from barium vacancies according to Heywang,
Jonker and Daniels, respectively. Below Curie temperature in tetragonal BaTiO
, barium
vacancies are neutral but above Curie temperature vacancies trap electrons and become
activated to singly ionised barium vacancies. Thus a negatively charged region forms at the
grain boundary and a positively charged space region forms adjacent to the boundaries. A
potential barrier is created between these two regions. Since trapped electrons are not
available for conduction, the electrical resistivity increases around Curie temperature
(Steele, 1991).

Fig. 3.2 The height of the potential barrier and the electrical resistivity-temperature curve
(Steele, 1991).
The height of the potential barrier, 
(shown in Fig.3.2) is given by

= A /  = A / C . (T-Tc) (3-1)
where A is a constant,  is permittivity and Tc is the Curie temperature. Above Tc, the
height of the potential barrier, 
increases with decreasing  (Steele, 1991). The electrical
resistivity,  is directly proportional to the height of the potential barrier according to
  exp ( 
/ k.T ) (3-2)
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

Thus the exponential increase of the electrical resistivity,  is
  exp A / C.k (1- Tc / T) (3-3)
The barrier height of BaTiO
ceramics increases with increasing porosity. There is an
exponential relationship between the resistance of PTCR ceramics and the barrier height. If
the ferroelectric compensation of the ceramics at the room temperature is the same, the
resistance of PTCR ceramics increases with increasing porosity. With the increase of
porosity, the contact between individual grains looks more dispersed, which is helpful to
diffusion of oxygen into the ceramic bulk and to oxidation of grain boundaries. A large
PTCR jump is only possible for the samples sintered in an oxidizing atmosphere. The
oxygen concentration at grain boundary is in direct proportion to PTCR effect of ceramics
(Zhang, 2004).
Sintering under O
atmosphere affects not only barrier height but the resistance and
capacitance of grain boundaries. The effect is due to variation of the adsorbed gases at the
grain boundaries. The diffusion coefficient of oxygen at grain boundary is much higher than
that in bulk, which implies that there is a gradient of adsorbed oxygen density in the surface
of the grain. Sintering under O
atmosphere results in increasing number of oxygen
acceptors at the grain boundary and increase in resistance. The barrier height also increases
with sintering (Jiang, 2003).
3.1 Barium titanate as PTCR thermistor
Barium titanate, which has a perovskite structure, is a common ferroelectric material with a
high dielectric constant, widely utilized to manufacture electronic components such as
mutilayer capacitors (MLCs), PTC thermistors, piezoelectric transducers, and a variety of
electro-optic devices. Dielectrics and insulators can be defined as materials with high
electrical resistivities. A good dielectric is, of course, necessarily a good insulator, but the
converse is by no means true (Wang 2002). Barium titanate (BaTiO
) is a ceramic material
with a band gap of 3.05 eV at room temperature. Since the energy gap of barium titanate is
large, it is an insulator when it is pure. It can be made n-type semiconductor when doped
with donor dopants such as La
and Nb
. Poly-crystalline n-type semiconducting barium
titanate (BaTiO
) exhibits a behaviour known as the positive temperature coefficient of
resistivity (PTCR) effect. The electrical resistivity of n-type semiconducting barium titanate
increases by several orders of magnitude near the ferroelectric Curie temperature (120 ◦C)
(Kim 2000). At the Curie temperature, barium titanate undergoes ferroelectric to paraelectric
transition (Kim 2004). This behaviour is indicated in Fig. 3.3 (Wang 2002). It has also been
reported that single crystals of barium titanate exhibit negative temperature co-efficient of
resistivity (NTCR) properties.
In fact, PTCR materials can be divided into four groups: polymer composites, ceramic
composites, V
compounds and BaTiO3-based compounds (BaSrTiO
, BaPbTiO
(Wang 2002). Theoretical models of the PTCR effect have been proposed by both Heywang
and Jonker. Both theories and experiments report that the effect is a grain-boundary related
phenomenon (Park 2002). The grain boundaries of barium titanate show a large electrical
resistance and PTCR results from the adsorbed oxygen at the grain boundaries. The porous
ceramics exhibit large PTCR effect. According to Heywang, PTCR effect can be explained in

Powder Metallurgy


Fig. 3.3 Typical resistivity behavior of a BaTiO3 –type PTCR material (Wang 2002).
terms of temperature-dependent Schottky-type grain boundary potential barriers. These
potential barriers result from charge accumulation due to grain boundary acceptor states
such as segregated acceptor ions or adsorbed oxygen at the grain boundaries.
When the pure BaTiO
is sintered in air, the defect reaction is as follows:
(g) = Oo
+ V
Because of the high oxygen pressure at grain boundary when samples are sintered in air,
this reaction prefers to happen at the grain boundary rather than in the grain lattice.The
neutral barium vacancy may be ionized by the electron that was introduced by the donor
+ 2e- = V
’’ (3-5)
+ V
= 2D˙
+ V
’’ + 2Oo
+ 1/2O
(g) (3-6)
In pure donor doped BaTiO
ceramics (the microstructure is given in Fig. 3.4), the capturing
of a free electron by neutral barium vacancy at the grain boundary during ferroelectric
phase transition would respond for the PTCR effect (Qi, 2002).
PTCR effect can only be observed when barium titanate is sintered in air or oxidizing
atmospheres. However, when it is sintered under vacuum or in reducing gas atmospheres,
magnitude of the resistivity jump is reduced critically(Kim 2004). While the samples were
annealed in reduced air (98% N
+ 2% H
), anion vacancies (oxygen vacancies) were
generated and electron compensation would occur to maintain the electrical neutrality,
increasing the conductivity of the ceramics. Heywang considered that no oxygen was
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

absorbed in the grain boundaries to form a depletion layer, and therefore that vacancy
compensation (Ba compensation) was suppressed and the potential barrier was not built, the
ceramic exhibited a poor PTCR effect. As the oxygen pressure increases, foreign ions are
increasingly compensated by the oxidation process (formation of barium vacancies), and
this Schottky barrier was formed between grains and grain boundaries to benefit the PTCR
effect (Chen 2007). At low temperatures resistance is low since resistance is dominated by
the grains. Above Curie temperature the resistivity of grain boundary increases PTC
resistance. The disappearance of the domain structure above the tetragonal-
cubic/ferroelectric-paraelectric transition temperature (Tc) results in the energy barrier
being raised at all the grain boundaries (Wegmann 2007). Positive temperature coefficient
(PTC) materials prepared from doped semiconducting barium titanate ceramics can be used
in various kind of electronic circuitry as a switching device or as a constant temperature
heater (Wang 2002).

Fig. 3.4 The microstructure of BaTiO3 based PTCR materials sintered at (a) 800C and (b)
These PTC materials prepared from doped semiconducting BaTiO
ceramics can be used in
various kinds of electronic circuitry as a switching device or as a constant temperature
heater. Other important application of a PTC thermistor is the measurement/detection/
control of temperature or parameters related to temperature. These PTC materials are
known to have the highest temperature coefficient of resistance among all sensor materials
available (Vijatovic et al., 2008).
4. Humidity/gas sensing
Even though the use of sensors is well established in process industries, agriculture,
medicine, and many other areas, the development of new sensing materials with high
sensing capabilities is proceeding at an unprecedented rate. Numerous materials have been
utilized for humidity sensing, of which the metal oxides that are physically and chemically
stable have been extensively investigated at both room and elevated temperatures. Sensors
based on changes in resistance and capacitance, as shown in Fig.4.1, are preferred owing to
their compact size, which could facilitate miniaturization required for electronic circuitry
(Pokhrel et al., 2003).

Powder Metallurgy

These PTC materials are known to have a high temperature coefficient of resistance around
Curie point and the ability of self-limiting, so they are as well come out to be a very useful
device for sensor applications. The first semiconductor oxide gas sensors were reported by
Seiyama et al. in 1962. Since then, there have been numerous studies concerning such oxide
semiconductors as SnO
, ZnO
, In
, TiO
, Fe
, HfO
, and BaSnO
. They are nowadays
widely used in the detection of gases. Semiconductor oxide gas sensors are extensively
studied in order to improve their sensing characteristics, i.e., sensitivity, selectivity, stability,
and response rate, to various kinds of gases and to meet the increasing needs of sensors in
complicated systems and under strict conditions. A trial-and-error method is still mainly
used in the development of new sensor materials to replace existing sensor materials (Park
2004). Metal-oxide-semiconductor (MOS) sensors operate on the basis of the modification of
electrical conductivity of metal oxide layers, resulting from the interactions between O
, O

and O
species and gas molecules to be detected (Kim 2005).
It has been known that absorption or desorption of a gas on the surface of a metal oxide
changes the conductivity of the material. The sensitivity of a surface to a gas can be as low as
parts per billion (ppb). It is highly desirable that metal oxide semiconductor sensors have a
large surface area, so as to adsorb as much of the target analyte as possible on the surface,
giving a stronger and more measurable response (especially at low concentrations).

Fig. 4.1 Capacitive and resistive humidity sensors (Electronic Circuits and Projects Forum).
For n-type semiconductors, the majority charge carriers are electrons, and upon interaction
with a reducing gas an increase in conductivity occurs. Conversely, an oxidising gas serves
to deplete the sensing layer of charge carrying electrons, resulting in a decrease in
conductivity (Fine 2010). Barium titanate is an n-type semiconductor which has humidity
and gas (such as reducing gases) sensing properties (Park 2004, Wagiran 2005).
In recent years, solid state humidity sensors have been widely used for the measurement
and control of humidity in an industrial or household environment. The relative humidity
(RH), which is the ratio of actual vapour pressure of water to the saturated vapour pressure
at a particular temperature, is commonly used to measure humidity. The sensor materials
used are in general polymeric or ceramic. Ceramic humidity sensors are superior in
performance to the polymeric type because of ist stability towards a variety of chemical
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

species, wide range of operating temperatures and fast response to the changes of humidity.
The principle of humidity sensing by ceramic sensors is changes of electrical conduction or
capacitance due to water chemisorption and/or capillary conduction in the pores. The
important ceramic materials used for humidity sensing are TiO
, Mg
, hematite,
, perovskites. Perovskites such as (Ba, Sr) TiO
are protonic ionic conductors which
have also been used for humidity sensing. In perovskite type oxides (ABO
), the atoms on
site A are susceptible to humidity (Agarwal and Sharma, 2002).
Perovskite oxides form a group of ceramics demonstrating a variety of interesting properties
and promising applications. Some perovskite oxides, which are n-type semiconductors, such
as doped SrTiO
), exhibit dependence of conductivity on humidity in the
atmosphere they are exposed to. It has been proposed that H
O molecules adsorb on the
surface, and the adsorbed H
O molecules can release electrons to the conduction band and
consequently increase the electronic conductivity.

Fig. 4.2 A schematic picture of modern resistive semiconductor gas sensor (Aalto University,
Research Group Gas sensors).
This kind of response of electrical conductivity to humidity changes has been extensively
investigated for application as humidity sensors. As the response is related to surface
adsorption, humidity sensors made using such materials generally exhibit fast response.
However, the electronic conductivity can also be influenced by other gaseous species as H
CO, to name a few, which can be adsorbed along with H
O. As a result, there can be
significant interference from other species, and such sensors do not exhibit the selectivity
required of a practical sensor (Wang and Virkar, 2004).
Gas sensors as shown in Fig.4.2, are mostly n-type semiconductors. They detect reducing
gases in air e.g leak detection in gas pipelines, indication of petrol vapour in filling stations
or for alcohol tests in exhaled air. The working temperature of sensors are 200-600°C. The

Powder Metallurgy

grains of the sintered body are covered by adsorbed oxygen. It withdraws electrons from the
bulk, forming O
ions at the surface. Charge carrier concentration in the grain volume
decreases, potential barrier is formed at the grain boundaries. The electrical conductance is
decreased by oxygen adsorbtion. The molecules of reducing gas interact with adsorbed
oxygen lowering potantial barrier, increasing conductance of the sensor as follows
2CO + O
- = 2CO
+ e- (3-7)
Numerous reducing gases like hydrogen, methane, CO, ethanol vapor, H
S can be
determined in air (Gründler, 2007).
For humidity sensing, ceramic materials have mainly been used in the form of porous
sintered bodies.Thus, controlling porosity and surface activity is of great importance in
determining the humidity-sensitive electrical properties of ceramic products (Kim and
Gong, 2005).
4.1 Barium titanate as humidity/gas sensor
Ceramic humidity sensors can be classified into the semiconducting and protonic (or ionic)
type according to their conduction mechanism. The semiconducting type humidity sensor,
operating at elevated temperature (300-400°C), was known to exchange charge carriers (e-
and h’) between the adsorbed water molecule and ceramic body. While the protonic
humidity sensor, usually operating at room temperature, employs protonic (H
) conduction
on the surface of the sample as a sensing mechanism. Both sensors require porous ceramic
body for the adsorption of water molecule (Hwang and Choi, 1997).
To make an improved humidity sensor, the microstructure and electrical conductivity of the
material should properly be controlled. In general, ceramic humidity sensors are made to be
porous and electrically resistive. In this respect, BaTiO
, as a protonic ceramic humidity
sensor, has advantage in establishing the microstructure-property relation since the
microstructure and conductivity of BaTiO
can be easily controlled by varying the amount
of donor dopant such as La and the processing conditions (Hwang and Choi, 1997).
Ceramic sensors are based on the adsorption of water molecules in the pores, and variations
of the structure of the adsorbed layer, as a function of humidity. The sensing mechanism of
barium titanate ceramic humidity sensors at room temperature can be explained using ionic
conduction model. When few water molecules are available at low humidity, they
chemisorb on grain surfaces of the ceramic to form hydroxyl groups as surface charge
carriers. When initial water molecules are adsorbed, each water molecule is hydrogen-
bonded to two hydroxyls, and the dominant surface charge carriers will be H
. When still
more water is adsorbed, clustering of the water molecules takes place, forming a liquid-like
multilayer film of hydrogen-bonded water molecules, where each water molecule is only
singly bonded to a hydroxyl group. Since dissociation of H
into H
O and H
energetically favorable in liquid water, the dominant charge carrier in high moisture
environment is H+. These processes result in progressively decreasing impedance of barium
titanate ceramic when exposed to increasing humidity (Yuk and Troczynski, 2003).
Nanocrystalline BaTiO
material has better humidity sensing properties than ceramic
, such as lower resistance and high sensitivity. However, from the application point
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

of view, one needs to further reduce the resistance of the resistive humidity sensor from 10

to 10
, to minimize the humidity hysteresis and to increase the sensitivity and
repeatability(Wang, 2000).
Since water is a polar molecule, the negatively charged oxygen is electrostatically attached
to the positively charged Ba
ions of the sensor material. The initial monolayer is
chemisorbed due to the formation of a chemical bond under the influence of high
electrostatic field between Ba
and oxygen of the water layer. This layer, once formed, is not
further affected by exposure to humidity. The irreversible reaction for the first layer can be
given as
+ H
O= Ba
- OH + H
Once the first layer is formed, subsequent layers of water molecules are physically adsorbed.
This physically adsorbed water molecules dissociate because of the high electric fields in the
chemisorbed water layer as given below:
O= H
+ OH (3-9)
The charge transport occurs when H
ions release a proton to neighbouring water
molecules which accepts it while releasing another proton and so on. This is known as
Grothuss’s chain reaction (Agarwal and Sharma, 2002).
5. Porous ceramics
Porous ceramics find nowadays a large variety of applications in several devices like filters,
acoustic absorbers, catalytic components, selective membranes, humidity sensors, among
others. There are two main kinds of porous ceramics: reticulate ceramics and foam ceramics.
The former consists of interconnected voids surrounded by a web of ceramic; the latter has
closed voids within a continuous ceramic matrix. Various chemical routes, besides the
conventional solid state reaction route, have been developed for obtaining porous ceramics
without the use of additives to avoid unwanted impurities in the final ceramic piece. An
attractive alternative route, due to its simplicity and low cost, is the polymeric precursor
technique based on the Pechini’s patent, that results in sinteractive homogeneous powders.
Solid state sintering usually results in polycrystalline bodies with pores in the 100–1000 nm
range. Graphite has also been used successfully for increasing pore volume in ceramics. In
conventional ceramic sintering, by heating pressed powders, a porous network is usually
formed in the spaces between the particles necks, porous stability being achieved by
controlling the size of the particles. An easy way of controlling pore sizes during solid state
sintering is to control particle sizes of the mixing powders as well as the sintering
temperature and time. A previous dilatometric analysis enables one to choose the suitable
sintering temperature and time (Cosentino, 2003).
5.1 The definition of porosity and porosity of ceramics
The ceramics are made up of ionic and covalent crystals. However, real crystals are not
perfect but contain imperfections that are classified according to their geometry and shape.
Vacancy is an example of point defect, dislocation is a line defect, grain boundary is a planar
defect (Barsoum 1997). Pores are important features for the ceramics because powder

Powder Metallurgy

metallurgy methods i.e mixing, milling, shaping and sintering, always result in some
amount of porosity. Fig.5.1 indicates atomic mechanisms occurring during sintering. The
macroscopic driving force operative during sintering is the reduction of the excess energy
associated with surfaces. This can happen by (1) reduction of the total surface area by an
increase in the average size of the particles, which leads to coarsening and/or the
elimination of solid/vapor interfaces and the creation of grain boundary area, followed by
grain growth, which leads to densification (Barsoum 1997).
Pores are three dimensional or volume defects along with cracks, precipitates and particles.
Pores generally contain some gas inside them. The reason for this is the employment of an
sintering atmosphere during processing. However, it is regarded as a vacuum for simplicity.
Pores might be present inside the matrix phase i.e isolated pores and surfaces or they might
be located at the grain boundaries and grain junctions.

Fig. 5.1 Different diffusion mechanisms involved in sintering. The grain boundary and bulk
diffusion (1, 2 and 5) to the neck contribute to densification. Evaporation-condensation (4)
and surface diffusion (3) do not contribute to densification (Chiang 1997).
Three quantitites relating to pores are important; the size of the pores, distribution of
porosity and total amount of porosity. The shape of the pores is less important. By knowing
the constituent phases of the ceramic, theoretical and actual densities can be used to obtain
porosity. Mercury intrusion porosimetry can be used to determine pore size and porosity
distribution (Carter 2007).
Pores are usually quite deleterious to the strength of ceramics not only because they reduce
the cross-sectional area over which the load is applied, but more importantly because they
act as stress concentrators (Barsoum 1997). First, they produce stress concentrations. Once
the stress reaches a critical level, a crack will form and propagate. Because ceramics possess
no plastic-deformation, once a crack is initiated (started), it propagates (grows) until fracture
occurs. Second, pores (i.e., their size, shape, and amount) reduce the strength of ceramics
because they reduce the cross-sectional areas over which a load can be applied and,
consequently, lower the stress that these materials can support (Jacobs 2001).
Porosity can be converted from the density of the ceramics, measured by Archimedes’
principle. The density can be calculated from the weight determinations as follows:
 = {m
/ [m
- ( m
– m
)]} x 
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

where m
is the dry weight of the samples, m
is the weight after dipping into water for 4h.
and fluid impregnation and m
is the weight while immersed in water. mw is the weight of
the wire used to suspend the samples in water, which is also measured in water. All weights
are in grams. 
is the density of water (in g cm
), which is 
= 1.0017-0.0002315T
with T is water temperature in °C. The theoretical density of BaTiO
is considered to be 6.0 g
(Zhang, 2004).
5.2 Different kinds of pore-forming agents for porous BaTiO3 fabrication
PTCR applications
Porous semiconducting BaTiO
can be prepared by the thermal decomposition of barium
titanyl oxalate BaTiO(C
O or by the incorporation of graphite, polyvinylalcohol
(PVA), polyvinylbutyral (PVB), borides, silicides, carbides, partially oxidized Ti powders
and potato-starch to BaTiO
. These BaTiO
ceramics indicate large PTCR effects. Oxygen can
be adsorbed at the grain boundaries due to the presence of pores. Porous ceramics are more
favorable to the formation surface of acceptor states compared to dense ceramics. Porous
thermistors show better heat resistance than dense ones and thus can be used as overcurrent
protectors in electric circuits. (Kim, 2002, 2003)
For PTCR applications, among these agents polymers, different kinds of starch and
metallic additives can be employed. The polymer additives include polyvinylalcohol
(PVA), polyvinylbutyral (PVB) and (PEG). The starch can be in the sort of potato or corn
starch. Partially oxidized titanium can be another choice to enhance porosity. Carbon
addition can also result in the porosity increase. For humidity sensing applications,
graphite and PMMA additions can be made to enhance porosity. The effects of different
groups of PFAs on the grain size, porosity, PTCR and sensing property of barium titanate
ceramics will be mentioned in the following paragraphs. Some of the examples of PFAs
are given below, the detailed data on all the PFAs will be presented in the following
5.2.1 The effect of PEG on PTCR properties
Polyethylene glycol (PEG) with amount ranging from 1 to 20 wt.% was added to BaTiO
The porosity and the grain size as shown in Fig. 5.2, of the n-BaTiO
ceramic containing 5
wt.% PEG is 10.4 % and 6.3 μm, respectively, while it is 25.2 % and 5.0 μm, respectively for
the sample containing 20 wt.% of PEG. The crystalline structure of the n-BaTiO
ceramics at
room temperature was tetragonal and independent on the PEG content. PTCR jump of
ceramics containing 1 and 20 wt.% of PEG was measured to be 2.84x10
and 6.76x10
respectively as shown in Fig.5.3 (Kim, 2004).
5.2.2 The effect of starch based agents on PTCR properties
The potato-starch is another pore-former used in the production of BaTiO
ceramics. 1-20
wt.% of potato-starch was employed to fabricate porous microstructure. The maximum
porosity and minimum grain size was obtained for the sample containing 20 wt.% of potato-
starch, which were 37.2% and 3.4 μm, respectively. Pore-former increased porosity and
decreased the grain sizes (Kim, 2002).

Powder Metallurgy


(a) (b)

(c) (d)
Fig. 5.2 Porous microstructures of barium titanate ceramics with the addition of (a) PEG
(Kim, 2004) , (b) TiO
(Kim et al, , 2002), (c) corn-starch (Kim, 2002) and (d) potato-starch
(Kim, 2002).
1-20 wt.% of corn-starch was added into BaTiO
to produce porosity. The cavities formed
due to the burning-out of corn-starch during sintering act as the sites of the pore generations
as shown in Fig.5.2. The maximum porosity and minimum grain size, which were 44.0% and
3.1 μm, respectively was measured for the sample containing highest amount of corn-starch.
The samples containing corn-starch had a tetragonal crystal structure. The addition of corn-
starch into Sb-doped BaTiO
ceramics results in fine microstructures (3.1-3.8 μm). The
porosity and pore size increase and the grain size slightly decreases with corn-starch. The
porosity increases from 7.2 to 44% with corn-starch content from 0 to 20wt.%. The grain size
changed from 4.8 to 3.1μm. These porous ceramics are advantageous to oxidize grain
boundaries and to produce surface acceptor states (Kim, 2002).
The PTCR jump of the Sb-doped BaTiO
ceramics with corn-starch is observed to be over 10

and 1–2 orders of magnitude higher than that of samples without corn-starch as given in
Fig.5.3. The donor concentration i.e. Sb and grain size decrease and the electrical barrier
height of grain boundaries and the porosity increase with corn-starch content. As a result,
room-temperature resistivity increases with pore-forming agent content. The grain
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications


(a) (b)

(c) (d)
Fig. 5.3 PTCR properties of barium titanate ceramics with the addition of (a) PEG (Kim,
2004), (b) TiO
(Kim et al, , 2002), (c) potato-starch (Kim, 2002) and (d) graphite (Park, 2004).

Powder Metallurgy

boundary resistivity contributes largely to the total resistivity. The electrical resistivity of
grain boundaries significantly depends on the porosity. It can be deduced from this study
that, adsorption of chemically adsorbed oxygen atoms at the grain-boundaries increases
potential barrier height. Thus electrical resistivity of grain boundaries increases contributing
to total resistivity and resulting in a PTCR effect (Kim, 2002).
The electrical resistivities of Sb doped BaTiO
ceramics were measured during heating
under different atmospheres. The high-temperature resistivity under O
was higher than
that under both air and reducing atmospheres of N
and H
. The low resistivities obtained at
high temperatures in reducing atmospheres were due to desorption of chemisorbed oxygen
at the grain boundaries, decreasing potential barrier height. Besides, N
and H
gases diffuse
along the grain boundaries and react with chemisorbed oxygen atoms thus chemisorbed
oxygen is consumed and conduction electrons are released. This leads to a decrease in
potential barrier and as a result, PTCR effects were degraded.
The room-temperature resistivity of porous samples increased with corn-starch content. The
room-temperature and high temperature (above Curie temperature) resistivity values were
measured to be 1.37x10
and 1.52x10
Ωcm, respectively for the sample containing 1 wt.% of
corn-starch. For the rest of the samples, high-temperature resistivities were well above 10 8
ohmcm. Besides, the resistivities of the grain boundaries increased with increasing
temperature. The room-temperature and high-temperature resistivities were measured to be
and 2.79x10
Ωcm for the sample containing 1 wt.% of potato-starch. PTCR jump
was determined to be 7.93x10
for the same composition. The grain boundary resistivity
jump was 1.22x10
for this sample (Kim, 2002).
5.2.3 The effect of graphite agents on PTCR properties
0.3-2.0 mol of C (as graphite powder) was added to Sb doped BaTiO
to produce porous
ceramics. As-sintered samples indicated tetragonal perovskite crystal structure, irrespective
of the added graphite content. The porosity of the ceramics increased with graphite addition
by an enhanced evolution of CO and CO
gases associated with the exothermic reactions.
The highest porosity was determined to be 18.3%. The grain growth occurred during
sintering due to exothermic reactions and maximum grain size obtained was 160.7 μm.
However, as the graphite content increased, the grain sizes decreased. This is because the
existing graphites act as barriers to inhibit the grain growth during sintering, reducing the
grain size (Park, 2004).
The addition of graphite into BaTiO
substantially improved PTCR characteristics. For the
sample containing highest graphite, the maximum resistivity above Curie temperature was
measured to be above 10
Ωcm and PTCR jump was above 1.2x10
. The maximum room-
temperature resistivity was measured for this sample, which was 1.35x10
Ωcm and PTCR
jump was 5.86x10
. The magnitude of the PTCR effect of samples containing graphite
increased by 1-2 orders compared to the sample without porosifier as shown in Fig. 5.3. For
the porous samples, fast response to overcurrent (below 20s) was achieved (Park, 2004).
5.2.4 The effect of metallic agents on PTCR properties
Partially oxidized Ti powders were used as pore-former in a study of Kim, Cho and Park. Ti
powders were heated at 600°C for 1h in flowing oxygen gas and were covered with TiO
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

(Ti) powders with various contents (0-9 vol%) were added to (Ba,Sr)TiO
in vacuum.
Vacuum-sintered ceramics were exposed to paste-baking treatment in air in order to make
oxygen adsorption possible at the grain boundaries. The porosity increased and grain size
decreased by the addition of TiO
(Ti) powders. The interfaces between TiO
(Ti) powder of a
large size and BaTiO
powder of a small size provide the sites of the pore generations,
resulting in porosity increase. 46.0% and 1.0μm are the porosity and grain size values,
respectively for the sample containing 9 vol% of TiO
(Ti). Small grain size may be due to a
decrease in the driving force for the grain boundary movement with TiO
(Ti) addition.
(Ti) addition and vacuum sintering (low oxygen content and small grain size)
transformed the tetragonal crystal structure of BaTiO
to cubic. (Ba,Sr)TiO
containing 5 vol% of TiO
(Ti) showed excellent PTCR characteristic, i.e. low room-
temperature resistivity (1.8x10
Ωcm) and the high ratio of maximum resistivity to
minimum resistivity (1.5x10
) as shown in Fig. 5.3 (Kim et al, , 2002).
Humidity/Gas sensing Applications
5.2.5 The effect of graphite agents on humidity sensing
The effect of graphite addition on grain size, porosity and humidity sensing properties was
investigated. Four different graphite containing compositions (C-1/C-4) were sintered at
1200-1500°C for 2-6h.
The graphite addition increased the grain size of the donor doped samples due to heat
generated by the exothermic reactions of carbon combustion. However, the grain size
decreased by further addition of graphite powder, because of the graphite segregation at the
grain boundaries (Ertug et al., 2005). This result is in good agreement with (Park, 2004)’s
study on PTCR properties.
The addition of graphite resulted in a porous microstructure as shown in Fig.5.4. The
maximum grain size was measured for the sample (BTLC-1) sintered at 1500°C for 6h. and it
was 15.7 μm. Upon sintering at 1500°C for 6h., the grain size of BTLC-4 composition reached
13.3 μm. The grain size decreased drastically when the sintering temperature was decreased
from 1500°C to 1200°C for 2-6h. For the sample sintered at 1200°C for 6h., the grain size was
measured to be 5.7 and 5 μm for BTLC-1 and BTLC-4 compositions, respectively (Ertug et
al., 2005).
The porosity percentage of the graphite containing compositions changed dramatically
when the sintering temperature was increased from 1200 to 1500°C. The porosity change
was slight between 1200-1400°C however after sintering at 1500°C, most of the porosity in
the microstructure was eliminated and dense samples were obtained (Ertug et al., 2005).
The minimum porosity percentage was measured for the sample (BTLC-1) sintered at
1500°C for 6h. and it was 12 %. When graphite addition was increased from 3.5 to 6.5%, the
porosity percentage changed from 59 to 64.9 %upon sintering at 1200°C for 6h (Ertug et al.,
The humidity sensing properties of graphite containing compositions were given in Fig.5.5
and 5.6. The transfer function of the humidity sensitive samples, which were sintered at
1200°C was of exponential type. At the low humidity percentages, the electrical resistance
decreased with relative humidity linearly. When the relative humidity increased beyond the

Powder Metallurgy


(a) (b)

(c) (d)
Fig. 5.4 The porous microstructures of graphite containing barium titanate samples (Ertug,
critical value, the electrical resistance-relative humidity curve became exponential. The
minimum electrical resistance was measured for the samples which contain higher porosity
percentages. When the content of the graphite was increased, the transfer function of the
humidity sensitive sample did not change, i.e exponentail curve. However, the electrical
resistance values indicated a slight decrease due to higher porosity percentage (Ertug, 2008).
When the samples were sintered at 1500°C, porosity percentages decreased by a great
amount. As shown in Fig.5.6, as-sintered samples did not exhibit a change in the electrical
resistance with relative humidity (Ertug, 2008).
5.2.6 The effect of PMMA agent on humidity sensing
The three different barium titanate compositions containing PMMA as pore forming agent
were investigated. The samples were sintered at 1200-1500°C for 2-6h. The grain size of the
samples decreased with PMMA content. The maximum grain size was measured for the
sample sintered at 1500°C for 6h. and it was 14.87 μm. The maximum porosity was determined
for the sample containing maximum PMMA amount, which was 53.5 %(Ertug, 2006).
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications


Fig. 5.5 The humidity sensing property of graphite containing samples sintered at 1200°C
(Ertug, 2008).

Fig. 5.6 The humidity sensing property of graphite containing samples sintered at 1500°C
(Ertug, 2008).
The transfer functions of the barium titanate samples changed with sintering temperature.
When the samples were sintered at 1200°C, the transfer function was exponential.
However, when the sintering temperature was increased to 1500°C, the transfer function
of the composition containing lower PMMA, included a dead band region where at low
relative humidity, the electrical resistance remained constant with humidity. The dead
band region was valid up to 40% of relative humidity. As shown in Fig.5.7., the
composition which contained higher amount of PMMA, did not show a certain dead band
region (Ertug, 2008).
0 20 40 60 80 100
Relative Humidity (%)

BTLC-1122 BTLC-4122
0 20 40 60 80 100
Relative Humidity (%)

BTLC-1152 BTLC-4152

Powder Metallurgy


Fig. 5.7 The humidity sensing property of PMMA containing samples sintered at 1200°C
(Ertug, 2008).

Fig. 5.8 The humidity sensing property of PMMA containing samples sintered at 1500°C
(Ertug, 2008).
However, the electrical resistance did not change very much with relative humidity up to
40% of relative humidity. A clear drop in the electrical resistance was observed only after
60% of relative humidity (Ertug, 2008).
6. Conclusion
n-type semiconducting barium titanate ceramics have applications as thermistors and
sensors. For barium titanate in order to exhibit PTCR and gas (or humidity) sensing
property, first donor doping by trivalent or pentavalent cations is necessary. The
0 20 40 60 80 100
Relative Humidity (%)

BTLP-1122 BTLP-3122
0 20 40 60 80 100
Relative Humidity (%)

BTLP-1152 BTLP-3152
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

improvement of porosity is another requirement for PTCR and sensor applications. Porous
barium titanate ceramics are generally fabricated using different kinds of pore-forming
agents (PFAs). The fabrication process is carried out using powder metallurgy methods
namely mixing with PFAs, shaping and sintering. PFAs, which are used for porous barium
titanate fabrication, can be grouped into several categories depending on the type of the
pore-former. Polymeric agents, metallic agents, starch-based agents and carbon-based
agents. An additional binder burn out step is employed for some of the pore-formers such as
PEG and direct sintering is done for other pore-formers such as starch and graphite. The
pore-forming mechanisms of several PFAs are different from each other. When oxidized Ti
powders are used as PFAs, the difference in the powder size of TiO
(18μm) and BaTiO

(1m) powders lead to pore generation. However, gas evolution during the burning of carbon
results in porosity when graphite PFA is used. When polymeric agents are used to produce
porosity, organic components are pyrolized upon binder removal process. The porosity
improvement by PFAs lead to humidity or reducing gas sensitivity of barium titanate based
ceramics. Besides, the oxidation of the grain boundaries are enhanced by porosity increase
which in turn improves PTCR property. Brand new PFAs such as several polymeric agents,
can be used to produce porous barium titanate or known to be effective PFAs can be applied
to it in order to produce more sensitive sensors or to obtain more pronounced PTCR effect in
. The effect of PFAs on the porosity (and pore size) and grain size of barium titanate
ceramics and also the different pore-forming mechanisms of PFAs are currently open to
further investigation since this particular area is based on trial and error method.
7. Acknowledgment
Dedicated to my beloved sister, Deniz for supporting me through my academic life.
I wish to thank my Ph.D. thesis advisor Prof.Dr. A. Okan Addemir at the Istanbul Technical
University, who enabled the beginning and the completion of the work on porous barium
titanate based humidity sensors.
I would like to express my gratitude to Assistant Prof. Dr. Tahsin Boyraz at Cumhuriyet
University, who gave me the idea of working on and writing a book chapter on powder
metallurgy. I truly appreciated all the time and advice he gave me throughout my time at
I also wish to thank my collegue at ITU, Gülten Sadullahoğlu, Ph.D. for her constant support
in my studies.
8. References
Agarwal, S. and Sharma, G.L. (2002). Humidity sensing Properties Of (Ba, Sr)TiO3 thin films
grown by hydrothermal-electrochemical method, Sensors and Actuators B, 85, 205-
Barsoum, M.W. (1997). Fundamentals Of Ceramics, The McGraw-Hill Companies, Inc., pp.149-
Brook, R.J. (1991). Concise Encyclopedia of Advanced Ceramic Materials, Pergamon Press.
Carter, C.B. and Norton, M.G. (2007). Ceramic Materials Science and Engineering, Springer
Science + Business Media LLC, pp. 277-279.

Powder Metallurgy

Chen, Y.C. (2007). Annealing Effects Of Semiconducting Barium-Titanate Thermistor, Journal
of Marine Science and Technology, Vol. 15, No. 4, pp. 307-314 .
Chiang, Y.M., Birnie, D. and Kingery, W.D. (1997). Physical Ceramics: Principles for Ceramic
Science and Engineering, John Wiley & Sons, Inc.
Cosentino, I.C., Muccillo, E.N.S. Muccillo, R. (2003). Development of zirconia-titania porous
ceramics for humidity sensors, Sensors and Actuators B, 96, pp. 677-683.
Desu, S.B. (1990). Interfacial Segregation in Perovskites: III, Microstructure and Electrical
Properties. J. Am.Ceram.Soc., 73, pp. 3407-15.
Ertug, B., Boyraz, T., Addemir, O., (2005). Effect of graphite content on the porosity
and microstructural characterization in La doped BaTiO
” ,IX Conference &
Exhibition of the European Ceramic Society, 19 - 23 June, Portoroz, Slovenia.
Ertug, B., (2008). The Effect of Pore-Forming Agents on the Electrical Properties of Barium
Titanate Based Ceramics Under Humid Environment, Istanbul Technical
University, Institute of Science and Technology.
Ertug, B., Boyraz, T., Addemir, O., (2006). PMMA Katkısının BaTiO
Esaslı Poroz
Seramiklerin Porozite ve Mikroyapısal Özellikleri Üzerine Etkisinin İncelenmesi,
Materials 2006, 11th International Materials Symposium, 19-21 April, Denizli-Turkey.
Fine, G.F., Cavanagh, L.M., Afonja, A. and Binions, R. (2010). Metal Oxide Semi-Conductor
Gas Sensors in Environmental Monitoring, Sensors, 10, pp. 5469-5502.
Gründler P. (2007). Chemical Sensors, An Introduction for Scientists and Engineers, Springer.
Hutagalung, S.D. (2011). Technical Ceramics; Ferroelectric Ceramics Lecture Notes, School
of Materials and Mineral Resources Engineering, Universiti Sains Malaysia.
Hwang, T.J. and Choi, G.M. (1997). Electrical characterization of pororus BaTiO3 using
impedance spectroscopy in humid condition, Sensors and Actuators B, 40, pp. 187-
Jacobs, J.A. and Kilduff, T.F. (2001). Engineering Materials Technology, Prentice-Hall, Inc. pp.
Jiang, S., Zhou, D., Gong, S. and Guan, X. (2003). Effect of heat-treatment under oxygen
atmosphere on the electrical properties of BaTiO3 thermistor, Microelectronic
Engineering, 66, pp. 896-903.
Kim, J.G., Cho, W.S. and Park, K. (2000). PTCR characteristics in porous (Ba,Sr)TiO3
ceramics produced by adding partially oxidized Ti powders, Materials Science and
Engineering B, 77, pp. 255-260.
Kim, J.G. (2002) Effect of atmosphere on the PTCR characteristics of porous, Sb-doped
BaTiO3 ceramics produced by adding potato starch. Journal of Materials Science
Letters, 21, pp. 1645-1647.
Kim, J.G. (2002). Synthesis of porous (Ba,Sr)TiO3 ceramics and PTCR characteristics,
Materials Chemistry and Physics, 78, pp. 154–159.
Kim, J.G., Cho, W.S., Park, K. (2002). Effect of reoxidation on the PTCR characteristics of
porous (Ba,Sr)TiO3, Materials Science and Engineering B, 94, pp. 149–154.
Kim, J.G. (2003). Synthesis of porous (Ba,Sr)TiO3 ceramics and PTCR characteristics,
Materials Chemistry and Physics, 78, pp. 154-159.
Kim, J.G., Tai, W.P., Kwon, Y.J., Lee, K.J. , Cho, W.S. , Cho, N.H., Whang, C.M. and Yoo, Y.C.
(2004). Effects of reducing and oxidizing atmospheres on the PTCR characteristics
of porous n-BaTiO
ceramics by adding polyethylene glycol, Journal of Materials
Science: Materials in Electronics, Volume 15, Number 12, pp. 807-811.
The Fabrication of Porous Barium Titanate Ceramics
via Pore-Forming Agents (PFAs) for Thermistor and Sensor Applications

Kim, J.G., Tai, W.P., Lee, K.J., Cho, W.S. (2004). Effect of polyethylene glycol on the
microstructure and PTCR characteristics of n-BaTiO3 ceramics, Ceramics
International, 30, pp. 2223-2227.
Kim, Y.S., Ha, S.C., Kim, K., Yang, H., Choi, S.Y. and Kim, Y.T., Park, J.T., Lee, C.H., Choi, J.,
Paek, J. and Lee, K. (2005). Room-temperature semiconductor gas sensor based on
nonstoichiometric tungsten oxide nanorod film, Applied Physics Letters, 86, pp.
Kim, D.U., Gong, M.S. (2005). Thick films of copper-titanate resistive humidity sensor,
Sensors and Actuators B: Chemical, Volume 110, Issue 2, pp. 321-326.
Morrison, F.D., Coats, A.M., Sinclair, D.C., West, A.R. (2001). Charge Compensation
Mechanisms in La-doped BaTiO3. Journal of Electroceramics, 6, pp. 219-232.
Park, K., Kim, J.G. and Cho, W.S. (2002) Effect of partially oxidized Ti powders on the
microstructure and PTCR characteristics of (Ba,Sr)TiO
, Journal of Materials Science:
Materials in Electronics, Volume 13, Number 1, pp. 13-19.
Park, K. (2004). Characteristics of porous BaTiO3-based PTC thermistors fabricated by
adding graphite powders, Materials Science and Engineering B, 107, pp. 19-26.
Park, K. (2004). Characteristics of porous BaTiO3-based PTC thermistors fabricated by
adding graphite powders, Materials Science and Engineering B, 107, pp. 19-26.
Pokhrel, S., Jeyaraj, B., Nagaraja, K.S. (2003) . Humidity-sensing properties of ZnCr2O4–ZnO
composites, Materials Letters, 57, pp. 3543–3548.
Qi, J., Gui, Z., Wang, Y., Zhu, Q., Wu, Y. and Li, L. (2002). The PTCR effect in BaTiO3
ceramics modified by donor dopant, Ceramics International, 28, pp. 141-143.
Smyth, D.M. (2000). The effects of dopants on the properties of metal oxides. Solid State
Ionics, 129, pp. 5-12.
Steele, B.C.H. (1991). Electronic Ceramics, Elsevier Applied Science, Chapman and Hall, New
Tsur Y., Dunbar T.D., Randall C.A. (2001). Crystal and Defect Chemistry of Rare Earth
Cations in BaTiO3. Journal of Electroceramics, 7, pp. 25-34.
Vijatovic, M.M., Bobic, J.D., Stojanovic, B.D. (2008). History and Challenges of Barium
Titanate: Part II. Science of Sintering, 40, pp. 235-244.
Wagiran, R., Wan Zaki, W.S., Mohd Noor, S.B., Shaari, A.H. and Ahmad, I. (2005).
Characterization Of Screen Printed BaTiO3 Thick Film Humidity Sensor,
International Journal of Engineering and Technology, Vol. 2, No. 1, pp. 22-26.
Wang, W. and Virkar, A.V. (2004). A conductimetric humidity sensor based on proton
conducting perovskite oxides, Sensors and Actuators B, 98, pp. 282-290.
Wang, J., Xu , B., Liu, G., Zhang, J., Zhang, T. (2000). Improvement of nanocrystalline BaTiO
humidity sensing properties, Sensors and Actuators B, 66, pp. 159-160.
Wang, H.L. (2002) Structure and Dielectric Properties of Perovskite-Barium Titanate
(BaTiO3), Submitted in Partial Fulfillment of Course Requirement for MatE 115,
San Jose State University.
Wegmann, M., Brönnimann, R., Clemens F. and Graule, T. (2007). Barium titanate-based
PTCR thermistor fibers: Processing and properties, Sensors and Actuators A: Physical,
Volume 135, Issue 2, pp. 394-404.
Yuk, J., Troczynski, T. (2003). Sol–gel BaTiO3 thin film for humidity sensors, Sensors and
Actuators B, 94, pp. 290-293.

Powder Metallurgy

Zhang, D., Zhou, D., Jiang, S., Wang, X. and Gong, S. (2004). Effects of porosity on the
electrical characteristics of current-limiting BaTiO3-based positive-temperature-
coefficient (PTC) ceramic thermistors coated with electroless nickel-phosphorus
electrode, Sensors and Actuators A, 112, pp. 94-100.
Zhou, L. (2000). Processing Effects on Core-Shell Grain Formation in ZrO2 Modified BaTiO3
Ceramics, Ph.D Thesis, University of Cincinati.
Springer Science+Business Media, Available from: http://www.springerimages.com/
Electronic Circuits and Projects Forum, Available from: http://www.electro-tech-
Aalto University, Department of Micro and Nanosciences, Research Groups Gas sensors,
Available from:
Hybrid Gas Atomization for Powder Production
Udo Fritsching and Volker Uhlenwinkel
University of Bremen
1. Introduction
Atomization of a molten liquid melt is a versatile method for powder production. Main
operational parameters and physical conditions for a suitable atomization process for
different applications in powder production are:
- melt type: surface tension, viscosity and temperature range (solidification temperature),
- process type: aimed throughput, energy efficiency,
- product type: particle size distribution (mean particle size and width of distribution).
Most atomization processes e.g. for metal powder production are gas atomization processes
using twin fluid atomizers. For these applications, besides the use of standard atomizer
types, relevant developments are ongoing to derive advanced atomizer concepts for
improved powder products and energy and resources effective atomization processes.
Main aims of atomizer developments are:
- minimization of (mean) particle sizes,
- narrowing of the particle size distribution width,
- effective use of resources (gas and feed stock),
- technical production of complex melt systems for powder applications.
In this contribution the development of gas atomization units is described, which utilize
characteristic flow effects of a molten metal stream in relation to the flow of compressed
gases for efficient fragmentation of liquids and melts. The interaction of gas and liquid melt
in the twin-fluid atomization process is used for understanding and optimization of melt
fragmentation processes.
2. Atomization of liquids and melts
Atomization of liquids (e.g., pure liquids, solutions, suspensions and emulsions, or melts) is
a classic process engineering unit operation. The process of liquid atomization has
applications in numerous industrial branches, for example, in chemical, mechanical,
aerospace, and civil engineering as well as in material science and technology and
metallurgy, food processing, pharmaceuticals, agriculture and forestry, environmental
protection, medicine, and others.
Within atomization, the bulk fluid (continuous liquid phase) is transformed into a spray
system (dispersed phase: droplets). The disintegration process itself is caused either by

Powder Metallurgy

intrinsic (e.g., potential) or extrinsic (e.g., kinetic) energy, where the liquid, which is
typically fed into the process in the form of a liquid jet or sheet, is atomized either due to the
kinetic energy contained in the liquid itself, by the interaction of the liquid sheet or jet with a
(high-velocity) gas, or by means of mechanical energy delivered externally, e.g., by rotating
The main purpose of technical atomization processes is the generation of a significantly
increased gas–liquid interface in the dispersed multiphase system. All transfer processes
across the gas–liquid phase boundary directly depend on the driving potential difference of
the exchange property (heat, mass, or momentum) and the size of the exchange surface. This
gas–liquid contact area in a dispersed spray system is correlated with the integral sum of the
surfaces of all individual droplets within the spray. An increase in the relative interphase
area in a dispersed system intensifies momentum, heat, and mass-transfer processes
between the gas and liquid. The total flux of exchange within spray systems is thereby
increased by several orders of magnitude.
Within the atomization process typically a liquid stream (jet or sheet) is disintegrated into a
large number of droplets of various sizes. As all of the solid state materials as minerals,
ceramics, glass and metals may be converted into the liquid state by melting, powders from
all of these materials may be produced via atomization of the liquid melt and consecutive
solidification of the droplets in the spray to achieve a powder.
The area of melt atomization differs in several ways and properties from a conventional
atomization process (lets say from water or fuel atomization) due to the physical melt
conditions as:
- high temperatures above the relevant melting point
- comparably high surface tensions (metal melts)
- comparably high viscosity (glass or ceramic melts)
The general physics, devices and application of (conventional) atomization and spray
processes have been reviewed and intensively published as e.g. in (Lefebvre, 1989; Bayvel &
Orzechowski, 1993; Fritsching; 2006; Ashgriz, 2011).
2.1 Gas atomization of liquid melts
Several principle atomisation mechanisms and devices exist for disintegration of molten
metals. An overview on molten metal atomisation techniques and devices is given e.g. by
(Lawley, 1992; Bauckhage, 1992; Yule & Dunkley, 1994, and Nasr et al., 2002). In the area of
metal powder production by atomisation of molten metals typically twin-fluid atomisation
by means of inert gases is used. Main reasons for using this specific atomisation technique
- the possibility of high throughputs and disintegration of high mass flow rates
- the greater amount of heat transfer between gas and particles for rapid partially cooling
the particles
- the direct delivery of kinetic energy to accelerate the particles towards the
substrate/deposit for compaction
- the minimization of oxidation risks of the atomised materials within the spray process
by use of inert gases.

Hybrid Gas Atomization for Powder Production

A common characteristic of the used variations of twin-fluid atomizers for molten metal
atomisation is the vertical exit of the melt jet from the tundish via the (in most cases
cylindrical) melt nozzle in the direction of gravity. Also in most cases the central melt jet
stream is surrounded by a gas flow from a single (slit) jet configuration or a set of discrete
gas jets, which are flowing in parallel direction to the melt flow or within an inclination
angle of attack towards the melt stream. The coaxial atomizer gas usually exits the atomizer
at high pressures with high kinetic energy.
Two main configurations and types of twin-fluid atomizers need to be distinguished within
molten metal atomisation. The first kind is called the confined or close-coupled atomizer
and the second kind is called the free-fall atomizer. Both concepts are illustrated in Figure 1.

Fig. 1. Gas atomization principles of melts: Close coupled atomizer (left) and free-fall
atomizer (right)
The gas flow in the close-coupled atomizer immediately covers the exiting melt jet. Within
the confined atomizer the distance between gas exit and melt stream is much smaller than in
the free-fall arrangement, where the melt jet moves a certain distance in the direction of
gravity before the gas flow impinges onto the central melt jet. The close-coupled
configuration generally tends to yield higher atomisation efficiencies (in terms of smaller
particles at identical energy consumption) due to the lower distance between the gas and
melt exits. But the confined atomizer type is more susceptible for freezing problems of the
melt at the nozzle tip. This effect is due to the extensive cooling of the melt by the expanding
gas flow, which exits in the close-coupled type nearby to the melt stream. During isentropic
gas expansion the atomisation gas temperature is lowered (sometimes well below 0°C).
Because of the close spatial coupling between gas and melt flow fields, this contributes to a
rapid cooling of the melt at the tip of the melt nozzle. The freezing problem is relevant
especially for spray forming applications. In this process, in all technical applications a
discontinuous batch operation is performed (e.g. due to the batchwise melt preparation or
the limited preform extend to be spray formed). The operational times of spray forming
processes are ranging from several minutes up to approximately one hour. The thermal
related freezing problem is most important in the initial phase of the process directly when
the melt stream exits the nozzle for the first time. At that point the nozzle tip is still cool and
needs to be heated first e.g. by the hot melt flow. This heating process lasts a certain time.
Therefore, thermal related freezing problems are often to be observed in the first few
seconds of a melt atomisation process.

Powder Metallurgy

In addition to the thermal related freezing problem within the melt nozzle, also chemical or
metallurgical related problems in melt delivery systems are found frequently. Not all of
these problems are solved yet in melt atomisation applications. A range of problems arises
from the possible change of material composition of the melt, or that of the tundish or
nozzle material, due to possible melt/tundish reactions or melt segregational effects from
diffusion. This reaction process kinetics is somewhat slower than the thermal freezing
process kinetics mentioned before and may contribute to operational problems at a later
temporal state of process operation.
Free-fall atomizers are much less problematic than close-coupled atomizers in terms of
thermal freezing processes, as the melt jet stream and the gas stream are well separated at
the exit of the melt from the delivery system (tundish exit). Therefore, the cooling of the melt
due to the cold gas occurs at a later position than within close-coupled atomizers. The free-
fall atomizer obeys as an additional advantage in the frame of spray forming the possibility
of controlled mechanical or pneumatic scanning and therefore oscillating the gas atomizer
with respect to one axis. Also concepts of spatial/temporal distribution of segments of gas
jets within the nozzle can be used for atomisation. By doing so, the free-fall atomizer gives
the operator an additional degree of freedom with respect to the control and regulation of
the mass flux distribution of droplets in the spray. This most important physical property of
the spray within other atomizer nozzle systems can (within a running process) only be
influenced by changing the atomizer gas pressure. By controlled scanning of the nozzle, the
mass flux can also be distributed over a certain area (necessary e.g. for flat product spray
3. Free-fall atomizer design for melts
The design of a conventional free-fall atomizer is simple, robust and very reliable. Thus the
free-fall atomizer concept has several applications in liquid processing. Free-fall atomizers
are applied e.g. in spray drying facilities, in powder production devices, or in spray forming
processes (Yule & Dunkley, 1994; Fritsching, 2008). Figure 1 shows a sketch of a
conventional external mixing free-fall atomizer. The design is based on the combination of
two gas nozzle systems, namely a primary and a secondary gas nozzle (Fritsching, &
Bauckhage, 1992). The secondary nozzle is the main atomization unit. The concentric gas jets
from the secondary nozzle impinge onto the central liquid jet that is disintegrated due to
instabilities from the shearing action of the secondary gas flow and its relative velocity
(Markus et al., 2002). The disintegration process of the liquid jet in the atomization zone is
located underneath the secondary gas nozzle (Lohner et al., 2003; Heck, 1998). Due to this
separation of the atomization area and the atomizer body, the free-fall atomizer is mainly
used for atomization of viscous or high temperature liquids and melts.
Due to the inclination of the secondary gas flow, underneath the secondary gas nozzle a
recirculation gas flow area may occur depending on the atomizer design and operation
conditions. If the intensity of the recirculation flow is reasonably high, liquid ligaments or
droplets are transported back to the atomizer body. Here, these sticking fragments may clog
the gas and/or liquid orifices and negatively influence or even stop the atomization process.
In common free-fall atomizer designs the recirculation gas flow is suppressed by the use of a
primary gas nozzle. The primary gas flow is coflowing to the liquid jet thereby guiding the
liquid into the atomization area without recirculation.

Hybrid Gas Atomization for Powder Production


primary gas nozzle
secondary gas nozzle
secondary gas flow
primary gas
liquid/melt supply
recirculation gas
atomization zone

Fig. 2. Main components and areas of a conventional free-fall atomizer (From Czisch &
Fritsching, 2008a, with permission)
During operation of the free-fall atomizer, the gas mass flow rates of both gas nozzle
systems are controlled by the primary and secondary gas pressures respectively (pressure
levels are given as of overpressure) (Czisch & Fritsching, 2004). Problems may occur during
atomizer operation if the pressure ratio between the primary and the secondary nozzle is
improperly adjusted and an intense recirculation flow is generated. The applicable
maximum gas pressure of the primary gas nozzle is limited by initial disturbances that may
be generated on the liquid jet before reaching the atomization area. Therefore, the secondary
gas pressure is limited also because the secondary gas mass flow rate determines the
necessary primary gas mass flow rate for prevention of recirculation. This coupling of the
two nozzle systems limits the applicability of the free-fall atomizer (Fritsching, 2004; Heck et
al., 2000; Bergmann et al., 2001).
To improve the operating conditions of the free-fall atomizer it is intended to increase the
gas pressures while the recirculation flow is still suppressed. In this case the recirculation
gas flow underneath the secondary gas nozzle must be prevented in a different way. It is
proposed here that a cylindrical ring device installed into the gas flow of the secondary gas
nozzle can improve the atomization performance. The applied device utilizes the Coanda
effect (Wille & Fernholz, 1965) and the injector principle to influence the local atomization
gas direction. Within the confined flow device (called here the Coanda-ring) an under-
pressure zone is generated which deflects the gas flow downwards and increases the
pressure difference between the inner zone and the ambient gas pressure. This downward
reflection of the gas stream effectively increases the gas entrainment from outside the main
flow and therefore increases the total gas mass flow rate. This effect increase becomes
important because the liquid jet interaction within the secondary nozzle is increased and the
recirculation gas flow is suppressed.
3.1 Numerical simulation of free-fall atomizer gas flow
The improvement of the atomizer design and operating conditions is based on numerical
investigations of the gas flow field of a conventional free-fall atomizer (primary and
secondary gas nozzle; see figure 2). Nitrogen is used througout this study as atomizer gas.
By applying a axisymmetric two-dimensional flow simulation it is assumed that the gas exit

Powder Metallurgy

configuration is that of a circular slit nozzle (Lohner, 2002; Lohner et al., 2005).
Compressibility (assuming ideal gas flow) as well as turbulence effects (using the
conventional k-eps turbulence model) are taken into account. The gas condition at the
nozzle exit is assumed as perfectly expanded (supersonic). Pressure levels are given as
overpressure (pressure level exceeding the ambient pressure). The orifice areas of primary
and secondary gas are kept constant. Therefore, the gas mass flow rates of the primary and
the secondary gas nozzle are varied by changing the stagnation gas pressures for primary
and secondary gas, respectively.
In Figure 3 the influence of the primary gas pressure, at constant secondary gas pressure (0.5
MPa), is shown in terms of gas streamlines. Two different operating conditions are to be
distinguished, separated by the center line in the figure. On the left of the figure the gas flow
field is illustrated with primary and secondary gas flow. One can see a reasonable
recirculation flow only close to the primary gas orifice. Entrained gas fom the outside as
well as the primary gas stream is flowing with the liquid through the passage of the
secondary gas nozzle. No recirculation gas flow underneath the secondary gas nozzle is
present. On the right of the figure the gas flow field is illustrated when only the secondary
gas (atomization gas) is applied but no primary gas. The shown streamlines illustrate an
intensive recirculation flow area located from underneath the secondary gas nozzle up to
the primary gas nozzle. Thus, at constant secondary gas pressure the intensity of the gas
recirculation underneath the secondary gas nozzle decreases with increasing primary gas
pressure. But the intensity of the recirculation gas flow close to the primary gas nozzle
increases with increasing primary gas pressure.

= 0.025 MPa
= 0.5 MPa
= 0 MPa
= 0.5 MPa

Fig. 3. Flow field of a conventional free-fall atomizer at different operating conditions (From
Czisch & Fritsching, 2008a, with permission)
In Figure 4 the axial gas flow velocity on the center line of the atomizer at constant
secondary gas pressure (0.5 MPa) is plotted for varying primary gas pressures. For primary
gas pressures exceeding 0.025 MPa the recirculation gas flow underneath the secondary gas
nozzle is suppressed (no negative velocities in that area). The intensity of the recirculation
close to the primary gas nozzle increases with increasing primary gas pressure.

Hybrid Gas Atomization for Powder Production

0 50 100 150 200
distance x [mm] .




0.000 MPa
0.009 Mpa
0.025 MPa
0.100 MPa
primary gas influence
secondary gas influence

Fig. 4. Axial gas velocity on the center line at varying primary gas pressures (secondary gas
pressure 0.5 MPa) (From Czisch & Fritsching, 2008a, with permission)
In Figure 5 the flow field of the free-fall atomizer without primary gas application is
illustrated as streamline distribution (left) and local static pressure distribution (right) at a
secondary pressure of 0.5 MPa. A recirculation flow field generated underneath the
secondary gas nozzle is visible. The area with highest pressure is located where the
atomization gas streams impinge onto the center line. The local static pressure exceeds 50
kPa in this area.

Fig. 5. Flow field (left) and pressure distribution (right) of a conventional free-fall atomizer
generated only by the secondary gas flow (scale indicates pressure level in Pa) (From Czisch
& Fritsching, 2008a, with permission)
Figure 6 shows the flow situation of a free-fall atomizer without primary gas application but
with the installed Coanda-flow ring device at a secondary gas pressure of 0.5 MPa. On the
left the flow field is illustrated by its streamline distribution and on the right the static
pressure distribution is displayed. No recirculation flow field is generated in this case. On
the right it can be seen that at the entrance of the installed Coanda-flow device an
underpressure area is generated. Depending on the negative pressure ratio, between the
underpressure area close to the atomization zone and the environmental gas pressure, the

Powder Metallurgy

gas flow is deflected downwards. Thus the mass flow rate increases though the liquid
passage of the secondary nozzle suppressing the recirculation completely. The maximum
underpressure within the Coanda-flow device exceeds –60 kPa for the given secondary gas

Fig. 6. Flow field (left) and pressure distribution (right) of a free-fall atomizer generated only
by the secondary gas flow but with installed Coanda-flow device (scale indicates pressure
level in Pa) (From Czisch & Fritsching, 2008a, with permission)
The axial velocity on the center line of the atomizer is plotted in Figure 7 versus the varying
secondary gas pressure with usage of the Coanda-flow device. In this investigation no
primary gas nozzle is used at all. There is almost no effect on the axial gas velocity with
increasing gas pressure until the gas passed the installed flow device. Up to this point the
flow is completely dominated by the entrained gas flow from outside. After the gas passed
the Coanda-flow device, the axial gas velocity increases with increasing secondary gas
pressure. Thus it is possible to achieve a recirculation free flow without applying any
primary gas by using the Coanda-flow device.
3.2 Atomization experiments
Model atomization experiments are used to verify the outlined improvements of the
operation of the free-fall atomizer with the installed Coanda-flow device. The main aim in
this part of the investigation is the proper adjustment of the gas flow rates (or pressures) to
ensure stable atomization conditions. Within these model experiments water is atomized by
air at room temperature at various pressures (overpressure above the ambient pressure). In
Figure 8 (left) a photograph of the atomization area is shown without the Coanda-flow
device at 0.5 MPa secondary gas pressure and without primary gas application. Obviously
an unstable atomization occurs where droplets are transported from the atomization area
upwards towards the atomizer body due to the generated recirculation flow. The instability
of the atomization is visible even as three-dimensional structure (asymmetric). When the
Coanda-flow device is installed (Figure 8 right) the atomization at 0.5 MPa secondary gas
pressure and without primary gas becomes stable, i.e., no major recirculation in the
atomization area occurs at all.

Hybrid Gas Atomization for Powder Production

0 50 100 150 200 250
distance x [mm] .




0.5 MPa
0.7 MPa
0.9 MPa
primary gas influence
secondary gas influence

Fig. 7. Axial gas velocity on the center line at varying secondary gas pressures with installed
Coanda-flow device (no primary gas nozzle) (From Czisch & Fritsching, 2008a, with

Fig. 8. Model experiments without (left) and with Coanda-flow device (right) (p
= 0.5
MPa, no primary gas, liquid mass flow rate 435 kg/h) (From Czisch & Fritsching, 2008a,
with permission)
The derived Coanda-flow device is adapted within a free-fall atomizer for powder
generation from viscous mineral melts. In this application, hot atomization gas is used to
suppress fibre formation during viscous melt atomization and to produce spherical
particles. Results for achieved particle size spectra have been reported in (Lohner, 2002;
Lohner et al., 2005). In Figure 9 the particle size distributions of two different atomization
runs are compared at similar operating conditions but with and without the installation of
the Coanda-flow device. A mineral melt (liquid melt, no solid particles included) is
atomized at a liquid mass flow rate of 300 kg/h. The liquid temperature is app. 1873 K and
the gas temperature about 1273 K. The viscosity of the melt at the atomization temperature
is 0.2 Pas. The secondary gas pressure used in these cases is similar (0.55 MPa and 0.58 MPa
respectively). In the case without the flow device, a primary gas flow is applied at a pressure
of 0.18 MPa. This relatively high primary gas pressure allows the suppression of major gas
recirculation, but the appearance of the spray structure indicated that some instabilities

Powder Metallurgy

from the primary gas flow seem to occur. A rather broad particle size distribution results
with an indication of a bimodal distribution shape resulting from atomization presumably
due to the primary and the secondary gas flows. With the installed Coanda-flow device, a
very stable atomizer operation was observed. The resulting mass median droplet diameter
(MMD) is app. 280 µm for the conventional free-fall atomizer and app. 140 µm for the free-
fall atomizer with the installed flow device. Therefore, the droplet diameter was halved
when compared to the conventional free-fall atomizer results.
0 100 200 300 400 500 600 700
particle size [µm]


conv. free-fall atomizer
free-fall atomizer with flow device

Fig. 9. Particle size distribution with and without applied Coanda-flow device for mineral
melt atomization (without Coanda-flow device: p
= 0.55 MPa, p
= 0.18 MPa; with
Coanda-flow device: p
= 0.58 MPa, no primary gas; liquid mass flow rate 300 kg/h, liquid
viscosity 0.2 Pas, gas temperature 1273 K, melt temperature 1873 K) (From Czisch &
Fritsching, 2008a, with permission)
4. Hybrid atomization: Rotary film formation plus gas jet desintegration
Viscous melts, for example glass melts or liquid slags, have a comparably high viscosity and
low surface tension. Powder production by gas atomization of such viscous melts is a
difficult task because of the rapid cooling of the generated ligaments during melt
fragmentation that may cause fibre formation instead of a particulate product. In addition,
the increasing temperature of the melt within the process increases the melt viscosity that on
the other side decreases the dynamic energy transfer rate between gas and liquid. Therefore,
the disintegration time-scale increases and the liquid fragments may be transported out of
the area where the most effective atomization occurs. Depending on this decreasing energy-
transfer rate, the resulting particle size increases and the efficiency of the atomization
process decreases.
By using external mixing gas atomizers with heated atomization gases, in principle a
particle product can be obtained (Lohner, 2002; Lohner et al., 2005; Czisch et al., 2003). The
efficiency of the atomization process decreases with increasing viscosity and the minimum
particle size is limited by the available energy input (Yule & Dunkley, 1994; Strauss, 1999;
Dunkley, 2001). Concepts for atomization of highly viscous liquids or melts have already
been developed (Pickering et al, 1985; Fraser & Dombrowski, 1962; Campanile & Azzopardi,

Hybrid Gas Atomization for Powder Production

2003). Certain limits of these existing technologies are to be seen when realizing an
appropriate (low) particle size or a low fibre-to-particle ratio at high melt-mass flow rates.
To increase the efficiency and to decrease the resulting particle size of an atomization
process, one can increase the specific surface energy before atomization (Lefebvre, 1980).
With respect to twin fluid (gas) atomization, the fragmentation must take place in a region
where the velocity difference between the atomization gas and the melt is highest. Thus the
melt must be transformed and transported into an area where the maximum velocity
difference occurs. To increase the specific surface energy before atomization, a rotating
device such as a disc can be used, obtaining high efficiency for transformation. To prevent
liquid accumulation effects at the rim of the disc and to decrease the film thickness, the
rotary device must operate in sheet-formation mode, generating a free-flowing liquid film
from the edge of the disc. The generated film is atomized subsequently by a high-speed gas
flow emerging from an external mixing gas atomizer. This configuration allows the
generation of a high velocity difference between the atomization gas and the melt film. To
prevent fibre formation, the atomization gas needs to be heated. The formation of the free-
flowing film is influenced by the local gas flow field. The generated gas flow field can be
used to support the transport of the free-flowing film into the most efficient atomization
zone. Thus the gas flow field has to be determined by an appropriate atomizer design. An
atomizer concept based on the discussed aspects is the prefilming hybrid atomizer.
4.1 The prefilming atomizer concept
The aim of the development is an atomizer design for highly viscous liquids and melts at
high throughputs that produce a considerably low droplet size. The prefilming hybrid
atomizer introduced here is a combination of a single-fluid rotary atomizer and an external
mixing twin-fluid atomizer. The advantage of this specific design is that the feed material is
first spread out by a spinning disc due to centrifugal forces, thereby increasing the initial
liquid surface prior to the gas atomization process. The rotary atomizer is operated in sheet
formation mode so that in the vicinity of the rotary disc a free prefilm is formed. This
prefilm is subsequently guided into the atomization zone of the external mixing atomizer by
means of intrinsic aerodynamic forces in the gas flow field. In Fig. 10 the principle of the
prefilming hybrid atomizer is illustrated. To ensure stable atomization, aerodynamic forces
are used to protect the atomizer body from coming into contact and being blocked by parts
of the liquid (melt). A sufficiently small gas mass flow rate from the inside of the atomizer
protects the atomizer body.
The flow field of gas and liquid of the prefilming hybrid atomizer is illustrated in Fig. 11.
Different local flow regimes need to be distinguished. The inner entrainment reaches the
atomization area through the atomizer liquid stream passage. The outer entrainment
reaches the atomization area directly, and the recirculation gas flow reaches the atomization
area from below the rotary disc. All flow regimes are initialized be the atomization gas. The
film movement mainly depends on the acting aerodynamic and inertial forces caused by the
momentum distribution of the inner entrainment flow, the recirculation gas flow
momentum, and the film momentum itself. The aim of the prefilming hybrid atomizer
design is to generate a maximum recirculation momentum where the liquid film is
transported close to the gas outlet.

Powder Metallurgy


viscous liquid
external mixing
rotary disk

Fig. 10. Sketch of the prefilming hybrid atomizer (From Czisch & Fritsching, 2008b, with
gas flow

Fig. 11. Gas flowfield and effective momentum transfer of the prefilming hybrid atomizer
(From Czisch & Fritsching, 2008b, with permission)
4.2 Numerical simulation
Numerical simulations are used to derive the suitable atomizer design for efficient viscous
melt atomization. The aim of the numerical simulations is to derive conditions where a
maximum gas momentum on the free flowing liquid film acts against the main axial
atomization gas direction. Because the highest velocity difference between gas and liquid
and therefore the most efficient atomization area is close to the atomization-gas outlet, the
recirculation momentum must be maximized to transport the liquid film towards the outlet
of the atomization gas. Within the gas flow simulations the axial momentum of the
recirculation gas flow is analyzed at the location of the free-flowing liquid film. In Figure 12
the recirculation momentum depending on the atomization gas outlet angle is shown. The
recirculation momentum starts at gas outlet angles above 15°. For outlet angles less than 15°
the inner entrainment momentum dominates the interaction with the film. At outlet angles
above 15° the recirculation momentum dominates the interaction and it increases with
increasing outlet angle. A reasonable high recirculation momentum is reached at outlet
angles between 45° to 50°. The atomization gas outlet angle for the design of the hybrid
atomizer has therefore been set to 45°.

Hybrid Gas Atomization for Powder Production


0 20 40 60
atomization gas angle [




p = 0,89bar
p = 1,78bar
p = 2,64bar

Fig. 12. Axial recirculation momentum on the free flowing film (From Czisch & Fritsching,
2008b, with permission)
| = 45

o = 60

h =10mm
D = 124mm
= 0

Fig. 13. The constructive parameters of the hybrid atomizer design (From Czisch &
Fritsching, 2008b, with permission)
In the same way as the gas outlet angle, all constructive parameters of the atomizer have
been determined separately by variation runs within the simulation. In Figure 13 the
resulting optimum parameter set for the prefilming hybrid atomizer is shown. The design is
based for the atomization of mineral melts at a temperature of 1873 K where the liquid
viscosity is 1 Pas and the melt mass flow rate is aimed at 300 kg/h. A gas jet spacing of 0 has
been chosen, therefore a slit nozzle is used as gas-flow exit geometry.
4.3 Model experiments
In Figures 14 and 15 experimental results of atomization of water and glycerol are
illustrated. The mass median droplet size in the spray measured by laser diffraction is
plotted against the air-to-liquid mass-flow ratio (ALR). The efficiency of the hybrid atomizer
is evaluated in comparison with a conventional free-fall atomizer (Lohner et al, 2003). The
free-fall atomizer is a common device for powder production (Bauckhage & Fritsching,
2000) and spray forming applications (Fritsching and Bauckhage, 2006). When low viscous

Powder Metallurgy

liquid as e.g. water is atomized, the efficiency of the hybrid atomizer is low. For a constant
liquid-mass flow rate, at identical ALR the hybrid atomizer produces coarser particle sizes.
In this case the increase of the specific surface energy before atomization is not an
advantage. It becomes advantageous only when the viscosity increases. Fig. 15 shows that at
constant ALR, the hybrid atomizer produces finer particles than a conventional atomizer for
viscous glycerol atomization. In this case, a mass-median particle size of less than 30 µm is
0 0,5 1 1,5 2 2,5
ALR [-] .








conv. 400kg/h
conv. 700kg/h
conv. 1000kg/h
hyb. 400kg/h
hyb. 700kg/h
hyb. 1000kg/h

Fig. 14. Experimental results for atomizing water by air (From Czisch & Fritsching, 2008b,
with permission)
0 0,5 1 1,5 2 2,5
ALR [-] .








conv. 395kg/h
conv. 710kg/h
conv. 1001kg/h
hyb. 414kg/h
hyb. 696kg/h
hyb. 1007kg/h

Fig. 15. Experimental results for atomizing glycerol by air (From Czisch & Fritsching, 2008b,
with permission)
4.4 Melt atomization experiments
For atomizing viscous melts by the hybrid atomizer concept, a pilot plant for hot-gas
atomization has been used (Lohner at al., 2005). The main part of the plant is a spray tower
(about 5.5 m in height). On top of the spray tower, the material is melted by an induction
heating device. As material to be atomized, a blast-furnace slag is used for the trials. The
melt temperature before atomization is measured as 1813-1843 K with a mean temperature
of 1826 K. Through the bottom pouring crucible, the melt flows out due to gravity at melt-
mass flow rates up to 300 kg/h. The hybrid atomizer located underneath the crucible is
operated using heated gas. Gas temperatures up to 1273 K and gas pressures up to 2 bar

Hybrid Gas Atomization for Powder Production

(relative overpressure against ambient pressure) have been realized within hybrid atomization
runs. Hot gas is produced by means of a discontinuous heat exchanger (cowper). The ceramic
bulk material inside the cowper is heated by a propane burner. After this heating process,
compressed gas or steam is blown through the cowper. The preset values of atomization gas
temperature and gas pressure are obtained by mixing the heated gas from the cowper with gas
at room temperature. The resulting melt droplets in the spray may be quenched and solidified
about 2 m below the atomization nozzle. The solidified powder is collected at the bottom of
the spray tower, the fine powder fraction is deposited in the cyclone.
Fig. 16 shows a Scanning-Electron-Micrograph of the sprayed powder from the mineral melt
atomization experiment using the hybrid atomization system. A particle fraction of the as
sprayed material in the size range from 110 to 350 µm is shown. Mostly spherical particles
are to be seen, the fraction of particles to the fraction of fibers in total is above 90 %. In Fig.
16 (right) particles size results for the atomization of this mineral slag melt at different
atomization gas temperatures and gas pressures (relative) are shown. The mass-median
particle size of the resulting powder is plotted versus the atomization gas temperature. The
gas temperature given in the graph is after expansion (a.e.).

-50 100 250 400 550 700 850 1000
atomization gas temperature T
*,a.e. [






p0 = 0.42bar

p0 = 0.77bar

p0 = 1.00bar

= 1.55bar

p0 = 1.98bar

Fig. 16. Micrograph of hot-gas atomized mineral melt particles (size fraction 110 – 350 µm)
and results for atomizing mineral melt at an angular disc speed of 1500/min (From Czisch
& Fritsching, 2008b, with permission)
The maximum gas temperature before expansion is 1273 K. With increasing gas temperature
and increasing atomization gas pressure the particle size decreases. For this set of
experimental conditions, the achieved minimum mass median particle size is 210 µm.
Compared with the model experiments and the material properties of glycerol the minimum
particle size is rather coarse. This results from the limited maximum atomization gas
temperature available in the facility. The achieved temperature is still too low in comparison
with the melt temperature. Before and within the disintegration process, the melt is cooled
down rapidly, so the viscosity increases and the disintegration efficiency is limited.
5. Pressure-gas-atomization
Pressure-gas-atomization is another disintegration process which can be subdivided in two
steps: a pre-filming and a gas atomization step. During the pre-filming step, a large molten

Powder Metallurgy

metal surface is generated to prepare an efficient second disintegration step by gas
atomization. In the following, the development, the principle, and the results of the
pressure-gas-atomization are described.
5.1 Principle
Pressure-gas-atomization became possible after the development of the pressure atomizer.
The idea of a pressure atomizer for molten metal was born in the 90`s by Sheikhaliev and
later published by Dunkley and Sheikhaliev (Dunkley & Sheikhaliev, 1995). Originally, the
authors called this atomization technique “Centrifugal Hydraulic Atomization”. Here, the
metal was melted in a crucible placed in a pressure vessel and the molten metal was
tangentially fed into a small swirl chamber where the molten metal starts to rotate.
Typically, on the centreline of the swirl chamber a gas core develops due to centrifugal
forces. Finally, at the lower end of the conical part of the swirl chamber a hole with a
diameter between 1 and 2 mm allows the melt to leave the swirl chamber as a thin film
creating a hollow cone inside the spray chamber (fig. 17). Normally, a pressure difference of
0.4 to 1.0 MPa between the pressure vessel and the spray chamber is sufficient to achieve a
hollow cone of a molten metal. Commonly, this principle is used for cold liquids as an
effective atomization system. For a long time, it was doubted if this principle can be applied
to molten metal because of their much higher surface tension and the nozzle heating was
another challenge. Finally, the molten melt film becomes instable and disintegrates to
ligaments and relatively large droplets (mass median diameter far above 100 µm).


Fig. 17. Schematic (above) and picture (below) of pressure atomization using pure tin
(Achelis, 2009).
The pressure-gas-atomization was developed to achieve smaller particles and a narrow size
distribution. It combines the pressure atomization as a pre-filming step and the gas
atomization. As the second step a gas-ring-atomizer is used to disintegrate the molten metal
film, ligaments, and any large droplets. This idea results in a patent published in 2002
(Uhlenwinkel, 2002) and initiated intensive investigations (Lagutkin et al., 2003b, 2004b;

Hybrid Gas Atomization for Powder Production

Achelis et al., 2006, 2008, 2009). The hollow cone film/spray is surrounded by the gas-ring-
atomizer. Due to the high velocity gas flow from the gas-ring-atomizer, the hollow cone
spray from the pre-filming process is atomized into small droplets. Simultaneously, the
particles are mainly directed to the centreline of the atomizer axis (Fig. 18).

Fig. 18. Schematic (left) and pressure-gas-atomization of pure tin (right), (Achelis et al.,
5.2 Pre-filming
A pressure difference of about 0.4 MPa is necessary to achieve a fully developed hollow cone if
pure tin is used. Fig. 19 shows the effect of the pressure on the cone angle. A fully developed
cone is already achieved with a pressure of 0.45 MPa but the cone angle Θ is still small (29°).
Higher pressure yields in a large cone angle as demonstrated in the figure as well.

Fig. 19. Effect of the pressure p
on the spray cone angle Θ during pressure-atomization of
pure tin (Achelis, 2009).
A high speed video camera was used to investigate the break-up of the liquid metal film.
The pictures in fig. 20 represent single images of different alloys. The image section shows
the area below the gas-ring-atomizer (not in use). The tin film is hidden by the gas-ring-
atomizer and only the ligaments of the oscillating film are seen below the gas ring nozzle.

Powder Metallurgy

Verfahrensprinzip: Druck-Zerstäubung
Sn SnCu30 AlSi12
Images from high speed video camera
Recording frequency; 63 000 frames/s
10 mm

Fig. 20. Break-up of liquid metal films using several alloys. Different film length and break-
up mechanism (Achelis et al., 2010).
From the ligaments multiple droplets are generated which are clearly visible below the gas-
ring-atomizer. The film break-up of the SnCu30 alloy is totally different. The film can be
recognized as the dark area with some bright reflection zones at the surface. The film length
is much longer compared to pure tin and the break-up mechanism is dominated by a
perforation of the film. The holes in the film grow quickly until only ligaments are left which
will break into multiple droplets as well. Finally, the last image (right) shows that the metal
film of the AlSi12 alloy is again longer. Bright areas on the surface are just reflections. The
spray cone is not fully developed which is indicated by the non-conical shape of the cone. A
higher pressure p
is necessary to achieve a fully developed hollow cone for this alloy.
5.3 Gas flow in the vicinity of the gas-ring-atomizer
The gas flow generated by the gas ring atomizer has been studied carefully for the design of
the atomizer. The gas-ring-atomizer can induce strong recirculation zones which can affect
the atomization and result in an instable process. Simple pressure measurements on the
centreline of the atomizer can help to understand the gas flow in the vicinity of the gas-ring-
atomizer. Fig. 21 shows the pressure on a tube surface located on the centreline using
different gas-ring-atomizer designs. The location and the intensity of the maximum pressure
indicate the risk of recirculation zones. The highest risk is given for the ring slit nozzle with
a maximum pressure value of 25 kPa just 10 mm below the gas-ring-atomizer. The other
atomizers have several discrete holes with different exit angles and same total exit areas of
18.5 mm². An angle of 0° represents the direction straight downwards.
The flow situation was also studied by using water as fluid instead of molten metal. To get
an interior view of the spray cone a laser light sheet was used to illuminate the central plane
of the spray cone (Fig. 22). White arrows exhibit the exit and direction of the gas jets.
Obviously, the ring slit nozzle attacks the water film from the pressure nozzle directly and
droplets come very close to the gas nozzle. In case of molten metal this will lead to an
instable process. This result corresponds to the pressure measurements in the figure before.
Therefore, discrete holes in the ring gas atomizer are preferred for pressure-gas-atomization.
An exit angle of 0° was chosen for the atomization of molten metal.

Hybrid Gas Atomization for Powder Production

nozzle exit angle +15°, exit area 18.5mm²
nozzle exit angle 0°, exit area 18.5mm²
nozzle exit angle -15°, exit area 18.5mm²
Ring nozzle with 0°, exit area 19.1mm²

Distance from the gas nozzle z [mm]

Fig. 21. Pressure on the surface of a tube placed on the centreline of the gas-ring-atomizer
indicates the gas flow field in the vicinity of the atomizer. The effect of different atomizer
designs is shown (Lagutkin et al., 2003a/b).
+15 0 -15 0 (ring slit)
Gas nozzle exit angle β in °
Central plane of the spray cone illuminated by a laser light sheet
Pressure gas atomization, fluid: water
Bright area = high amount of particles

Fig. 22. Influence of different atomizer designs on the spray cone of a pressure-gas-atomizer.
The central plane of the spray cone is illuminated by a laser light sheet and water is used as
a liquid (Uhlenwinkel et al., 2003).
5.4 Melt flow rate
The production rate is an important feature for an atomization processes. Typically the mass
flow of the pressure gas atomization can be varied between 70 and 200 kg/h using tin and
tin-copper alloys. As expected, the melt flow rate can be adjusted by the pressure difference
between the pressure vessel and the spray chamber. The gas pressure in the gas-ring-
atomizer does not influence the melt flow.

Powder Metallurgy

5.5 Particle size
Typically, the mass median diameter without gas atomization (only pressure-atomizer) are
in the range of 400 to 500 µm. The gas to metal ratio (GMR, here in kg gas / kg melt) is the
key parameter to control the mean particle size. Fig. 23 shows that already with low GMR
(below 1.0) the mass median diameter can be varied between 40 and 100 µm for pure tin.
Even mean particles sizes of approximately 20 µm can be achieved. An increase of the
copper content up to 50 weight percent does not make a big difference to the mass median
diameter. Only a copper content of 63 weight percent results in larger mean particle sizes. It
is assumed the higher surface tension is the main reason for this behavior. All
measurements were accomplished by laser diffraction. A superheat of 100 K in the crucible
was enough to achieve mostly spherical particles.

Gas to metal ratio (GMR)

Fig. 23. Effect of the GMR on the mass median diameter for different alloys. The superheat
was almost 100K above the liquidus temperature of the alloy (Achelis et al., 2009).
Normally, the size distribution is characterized by another parameter, for example the
geometric standard deviation σ
= d
= d
. Both values are the same if the
distribution follows a log-normal size distribution. Here, the standard deviation is defined
as σ
= d
. The results in fig. 24 are plotted versus the mass median diameter.
Obviously, the geometric standard deviation depends on the mean particle size and a low
mean particle size results in a higher geometric standard deviation. This holds for pure tin
and this tin-copper alloys as well. A standard deviation below 2.0 achieved by a gas
atomization technique is considered extremely good.
Of course, there are other parameters effecting the particle size distribution, specially the
design of the atomizers. But the examples shown are representative. The mass median
diameter can be calculated by the following semi-empirical equation (Achelis, 2009):

37 . 0
5 . 0
4 . 0
1 . 0 6 . 0
2 3 , 50
1 01 . 0 1 07 . 0 45 . 1 |

O (







m d

µ o
o q

Here, the mass median diameter d
(in m) depends on the surface tension σ
, density ρ

and viscosity η
of the melt, the gas density ρ
, the gas velocity u
at a distance of 25 mm

Hybrid Gas Atomization for Powder Production


Fig. 24. Geometric standard deviation σ
versus the mass median diameter d
for different
alloys to characterize the particle size distribution of a pressure-gas-atomization (Achelis et
al., 2010).
from the orifice, the film thickness δ, the melt and gas mass flow


, and the spray
cone angle Θ. The film thickness is calculated from the equation given in (Rizk & Lefebvre,
1985 cited in Achelis, 2009) for a pressure atomizer:



66 . 3

In equation (2) D
means the exit diameter of the nozzle.
5.6 Particle velocity
Particle velocities are of interest for the verification of process modeling and for the
calculation of cooling and solidification rates. Several measuring techniques are available to
measure the particle velocity in the spray cone (e.g. Laser Doppler Anemometry LDA, Phase
Doppler Anemometry PDA, Particle Image Velocimetry PIV). Some results of a PDA
measurement are presented in fig. 25. Here, the mean velocity on the centreline of the spray
cone is plotted versus the gas pressure in the gas-ring-atomizer at a distance of 165 mm. The
molten tin was superheated to a temperature of 275 °C and the melt flow reached 180 kg/h.
The variation of the gas pressure resulted in a gas mass flow range between 117 and 198
kg/h. Because of the increased gas flow the mean particle size was reduced and both higher
gas velocities and smaller particles added up to mean particle velocity between 70 and 135
5.7 Particle shape
Generally, inert gas atomized metal show spherical shape. Therefore, the mean circularity C
= 4π projection area / (perimeter)² of the particles is close to 1.0. But frequent collisions
between liquid or semi-liquid droplets and solidified particles in the spray cone can result in

Powder Metallurgy

Pressure in the gas nozzle [MPa]

Fig. 25. Correlation between pressure in the gas nozzle and mean particle velocity on the
centreline for pure tin measured by Phase-Doppler-Anemometry (Lagutkine et al., 2004a)
agglomerations and/or multiple small particles (satellites) attached to a bigger one. This
will reduce the flowability of the powder which is an important quality feature. Often,
satellites occur because of the gas flow in the spray chamber. Close to the wall of the spray
chamber the gas velocity direction is contrary to the main flow direction and small particles
are transported to the spray cone again and collide with the droplets. This recirculation zone
inside the spray chamber can be avoided by a recirculation of the clean gas (after the particle
collector (cyclone and/or filter) and the fan). The clean gas is added at the top of the spray
chamber and - as a side effect – leads to a better visibility of the atomization process.
There appears to be a significant influence of the gas recirculation on the circularity of the
powder (figure 26). With gas recirculation the mean circularity of the powder is very close to
Mass median diameter d
with gas recirculation
without gas recirculation

Fig. 26. Mean circularity versus mass median diameter with/without gas recirculation
measured by image analysis (G3 Morphology, Malvern), (Achelis et al., 2010)

Hybrid Gas Atomization for Powder Production

1.0 and the SEM picture on the right side confirms this result clearly. Without gas
recirculation the circularity drops considerably and SEM figure shows the reason of this
behavior in agglomerates and multiple satellites.
6. Conclusions and future Investigations
Gas atomization is the most versatile method to produce powders by atomization of a liquid
melt. In this context, free-fall atomizers are external mixing gas atomizers based on a
combination of a primary and a secondary gas nozzle. The primary gas stabilizes the
atomization process and the secondary gas is used for atomization of the liquid. If only the
secondary gas is used, a gas recirculation flow underneath the secondary nozzle occurs. This
recirculation flow may cause the transport of atomized ligaments or droplets in the direction
of the atomizer body. Liquid impinging on the atomizer body may clog the gas orifices and
badly influence the atomizer performance. Thus in common free-fall atomizer design, the
primary gas is used to suppress the recirculation flow and to stabilize the atomization
To improve the operating conditions and to overcome limitations of the free-fall atomizer, a
Coanda-flow ring device was developed and installed inside the secondary gas flow. By
using the Coanda-flow device, the secondary gas flow is deflected in the downward
direction, the entrainment mass flow is increased, and, therefore, a primary gas nozzle is not
necessary for suppression of gas recirculation. The concept was derived and verified by
means of numerical gas flow simulations and model experiments. The application of the
flow-adapted design option for free-fall atomizers was demonstrated by the atomization of a
viscous melt. The results indicated a stable operation of the free-fall atomizer even at higher
secondary gas pressures, as well as a finer particle yield.
The concept of a prefilming hybrid atomizer has been introduced. The atomizer is designed for
atomizing highly viscous liquids and melts. Physical gas flow effects are utilized to increase
the atomizer efficiency. The hybrid atomizer design is based on numerical flow simulations of
the gas phase. Experiments where water a viscous model liquid (glycerol) a have been
atomized with air show a higher disintegration efficiency than a conventional external mixing
atomizer for the viscous liquid atomization process. Model experiments atomizing fluids of
low viscosity like water show an insufficient atomization efficiency of the hybrid atomizer
design. Thus the hybrid atomizer concept is a novel device for highly viscous liquids and
melts. The experimental results for atomization of a mineral melt by heated gases underline
the success of the hybrid atomization process with some limitations. For the present
conditions, a minimum mass-median particle size of 210 µm has been achieved. The coarse
product obtained by melt atomization is due to limitations of the available maximum
atomization gas temperature and may be overcome in future developments.
The complexity of the pressure-gas-atomization is a drawback for frequent applications in
industry. To overcome the problem the system can be simplified in the future if it is used
with low melting point alloys. Today, pumps for molten metal like tin alloys are available
and the complexity of the system can be reduced considerably because a pressure vessel is
not necessary anymore and thus the process principally can operate continuously.
Another future objective is the use of the pressure-gas-atomizer for coating applications. It
has already been proofed that this atomization system is dedicated to generate thick

Powder Metallurgy

coatings on a tube. As an example a steel tube with a diameter of the 80 mm was coated
with a 10 mm thick tin layer. The density of the coating was much better the than 99% of the
full density (Uhlenwinkel et al., 2008).
7. Acknowledgments
The results that have been presented here are mainly based on the project work of several
PhD students at the Particles and Process Engineering department of the University of
Bremen. Namely the contributions of Dr.-Ing. L. Achelis, Dr.-Ing. S. Markus, Dr.-Ing. H.
Lohner, Dr.-Ing. U. Heck, Dr. S. Pulbere and Dr.-Ing. C. Czisch are acknowledged. The
cooperation in these fields with international colleagues, namely Dr. S. Lagutkin, Prof. S.
Sheikhaliev and Dr. V. Srivastava is greatfully acknowledged.
The financial support of the projects has been mainly given by the German Research
Foundation (DFG), whose support during the past decade in several areas and through
projects is greatfully acknowledged.
8. References
Achelis, L.; Uhlenwinkel, V.; Lagutkin S. & Sheikhaliev, S. (2006): Atomization Using a
Pressure-Gas-Atomizer. Proceedings of the International Conference on Powder
Metallurgy 2006, Busan (South Koria), Sept. 2006
Achelis, L. & Uhlenwinkel, V. (2008): Characterisation of metal powders generated by a
pressure-gas-atomizer. MSE A 477 (2008), pp.15-20
Achelis L. (2009): Kombinierte Drall-Druck-Gaszerstäubung von Metallschmelzen. Shaker Verlag,
Aachen 2009, ISBN 978-3-8322-8012-3
Achelis, L.; Sulatycki, K.; Uhlenwinkel, V. & Mädler, L. (2010): Spray angle and particle size
in the pressure gas atomization of tin and tin-copper alloys, Proceeding of the
International Conference on Powder Metallurgy 2010, Florence (Italy), October 2010
Anderson, I.E. & Figliola, R.S. (1998). Observations of gas atomization process Dynamics, in
Modern Developments in Powder Metallurgy, Gummeson, P.U. &Gustafson, D.A. Eds.,
Metal Powder Industries Federation, Princeton, N.J., Vol. 20, pp. 205–223
Ashgriz, N. (2011). Handbook of Atomization and Sprays, Springer, New York
Bayvel, L. & Orzechowski, Z. (1993). Liquid Atomization, Taylor & Francis ,Washington
Bauckhage, K. (1992). Das Zerstäuben metallischer Schmelzen, Chem.-Ing.-Tech., Vol. 64 No.
4, pp. 322–332
Bauckhage, K. & Fritsching U. (2000). in: K.P. Cooper, I.E. Anderson, S.D. Ridder, F.S.
Biancaniello (Eds.) Liquid Metal Atomization: Fundamentals and Practice, TMS,
Warrendale, USA, pp. 23 – 36
Bergmann, D.; Fritsching, U. & Bauckhage, K. (2001) Simulation of molten metal droplet
sprays, Comp. Fluid Dynamics-Journal, Vol. 9, pp. 203 – 211
Bradley, D. (1973). On the atomization of liquids by high-velocity gases, Part 1, J. Phys. D:
Appl. Phys., Vol. 6, 1724–1736, Part 2, J. Phys. D: Appl. Phys., Vol. 6, pp. 2267–2272
Campanile, F. & Azzopardi, B.J. (2003). in: Cavaliere, A. (Ed) CD-ROM Proc. International
Conference on Liquid Atomization and Spray Systems ICLASS 2003, Sorrento, Italy, 13-
17.07.2003, ILASS-Europe

Hybrid Gas Atomization for Powder Production

Czisch, C.; Lohner, H.; Fritsching, U., Bauckhage, K. & Edlinger, A. (2003). in: K. Bauckhage,
U. Fritsching, J. Ziesenis, A. Uhlenwinkel, A. Leatham (Eds), Proc. Spray Deposition
and Melt Atomization Conf. SDMA 2003, Bremen, 22.-25.6.2003
Czisch, C.; Lohner, H. & Fritsching, U. (2004). Einfluss der Gasdüsenanordnung auf den
Desintegationsvorgang und das Zerstäubungsergebnis bei der Zweistoff-
zerstäubung, Chem.-Ing.-Tech., Vol. 76, No. 6, pp. 754-757
Czisch, C. & Fritsching, U. (2008a). Flow-adapted design for Free-fall atomizers, Atomization
and Sprays, Vol. 18 No. 6, pp. 511-522
Czisch, C. & Fritsching, U. (2008b). Atomizer design for viscous-melt atomization, Mat. Sci.
Engng. A Vol. 477 No. 1-2, pp. 21-25
Dombrowski, N. & Johns, W.R. (1963). Chem. Eng. Sci., Vol. 3, pp. 203-214
Dunkley, J.J. & Sheikhaliev, S. (1995): Single Fluid Atomization of Liquid Metals, Proceedings
of the International Conference on Powder Metallurgy & Particulate Materials, Seattle
(USA), May 1995, Vol. 1, pp. 79-87
Dunkley J.J. (2001). in: 2001 International Conference on Powder Metallurgy and Particulate
Materials PM
TEC 01, 2-29-2-35, 2001, Metal Powder Industries Federation,
Princeton, USA
Fraser, R.P.; Dombrowski, N. & Routley, J.H. (1962). Chem. Eng. Sci., Vol. 18, pp. 339-353
Fritsching, U. & Bauckhage, K. (1992). Investigations on the atomization of molten metals:
The coaxial jet and the gas flow in the nozzle near field, PHOENICS J. Comp. Fluid
Dynamics, Vol. 5, No. 1, pp. 81-98
Fritsching, U. (2004). Spray Simulation: Modeling and Numerical Simulation of Sprayforming
Metals, Cambridge University Press, Cambridge, UK
Fritsching, U. & Bauckhage, K. (2006a). Sprayforming of Metals, in: Ullmann´s Encyclopedia
of Industrial Chemistry, 7
Edition, 2006 Electronic Release, Wiley VCH,
Weinheim, Germany
Fritsching, U. (2006b). Spray Systems, in: Multiphase Flow Handbook, Ed.: C.T. Crowe,
CRC-Press, Boca Raton, Fl, USA, 2006, Chapter 8, pp. 8.1 - 8.100; ISBN: 0849312809
Heck, U. (1998). Zur Zerstäubung in Freifalldüsen, VDI Verlag GmbH Düsseldorf, Germany
Heck, U.; Fritsching, U. & Bauckhage, K. (2000). Gas-Flow Effects on Twin-Fluid
Atomization of Liquid Metals, Atomization and Sprays, Vol. 10, No. 1, pp. 25 – 46
Lagutkin, S.; Achelis, L.; Sheikaliev, S.; Uhlenwinkel, V. & Srivastava V. (2003a):
Atomisation Process for Metal Powder. Proceedings of the International Conference on
Spray Deposition and Melt Atomization 2003, Bremen (Germany), June, 2003,
Lagutkin, S. (2003b): Development of Technology and Equipment for Metal Powder Production by
Centrifugal-Gas Atomization of Melt. Ph.D. thesis. Ekaterinburg, Ural Department of
Academy of Sciences. 2003, in Russian
Lagutkin, S.; Uhlenwinkel, V.; Achelis, L.; Pulbere, S. &. Sheikhaliev S. (2004a): Centrifugal-
Gas Atomization: Preliminary Investigation of the Method, Proc. International
Conference on Powder Metallurgy 2004, Wien, Austria, Oct. 2004, Vol. 1, S.71-76
Lagutkin, S.; Achelis, L.; Sheikhaliev, S.; Uhlenwinkel, V. & Srivastava V. (2004b):
Atomization process for metal powder. MSE-A 383 (1), (2004), pp. 1-6
Lavernia, E.J. & Wu, Y. (1996). Spray Atomization and Deposition, J. Wiley & Sons, Chichester
Lefebvre, A.H. (1980). Prog. Energy Combust. Sci., Vol. 6, pp. 233-261
Lefebvre, A.H. (1989). Atomization and Sprays, Hemisphere Publ. Corp., US

Powder Metallurgy

Liu, H. (2000) Science and Engineering of Droplets: Fundamentals and Applications, William
Andrew Publ., Norwich, USA
Lohner, H. (2002). Zerstäuben von Mineralschmelzen mit Heißgas, PhD. thesis, University
Lohner, H.; Czisch, C. & Fritsching, U. (2003). Impact of Gas Nozzle Arrangement on the
Flow Field of a Twin Fluid Atomizer with External Mixing, Int. Conf. on Liquid
Atomization and Spray Systems, Sorrento, Sep. 2003 (Italien)
Lohner, H.; Czisch, C.; Schreckenberg, P.; Fritsching, U. & Bauckhage, K. (2005).
Atomization of viscous melts, Atomization and Sprays, Vol. 15, No. 2, pp. 169-180
Markus, S.; Fritsching, U. & Bauckhage, K. (2002). Jet Break Up of Liquid Metals in Twin
Fluid Atomization, Materials Sci. & Engng. A, No. 326, pp. 122 – 133
Pickering, S.J.; Hay, N.; Roylance, T.F. & Thomas, G.H. (1985). Ironmaking and Steelmaking,
Vol. 12 No. 1
Rizk, N.K. & Lefebvre A.H. (1985) : Internal Flow Characteristics of Simplex Swirl Atomizer.
Journal of propulsion and power, Vol. 1 (1985) No. 3, New York, pp. 193-199
Strauss, J.T. (1999). Hotter gas increases atomization efficiency, Metal Powder Report, Vol. 11,
pp. 24-28
Uhlenwinkel, V. & Sheikhaliev S., et al. (2002): Verfahren und Vorrichtung zum Herstellen von
Metallpulver und keramischem Pulver, Pat.Nr. 10237213, 14.Aug. 2002
Uhlenwinkel, V.; Achelis, L.; Sheichaliev, S. & Lagutkin, S. (2003): A new Technique for
Molten Metal Atomization. Proc. ICLASS 2003, Sorento, Italy, June, 2003
Uhlenwinkel, V.; Achelis, L. & Mädler, L. (2008): New atomization process to achieve high
production rates during thermal spraying of thick coatings, Proceedings of the
International Thermal Spray Conference 2008, Maastricht (Netherlands), June 2008
Wille, R. Fernholz, H. (1965). Report on the First European Mechanics Colloquium on the
Coanda Effect, J. Fluid Mech., Vol. 23, pp. 801-819
Yule, A.J. & Dunkley J.J. (1994). Atomization of Melts, Clarendon Press Oxford, UK

Sign up to vote on this title
UsefulNot useful